Events

## Group Meetings

All group meetings take place in the Multimedia Signal Processing Lab, KEB 2211.

### Spring 2017

Group Meeting Schedule

### Fall 2016

Date: 09/13/2016
Speaker: Shawn Ma
Title:
Date: 09/20/2016
Speaker: Zhung-Han Wu
Title:
Date: 10/4/2016
Speaker: Qinyi Xu
Title:
Date: 10/11/2016
Speaker: Yi Han
Title:
Date: 10/18/2016
Speaker: Feng Zhang
Title:
Date: 10/25/2016
Speaker: Chen Chen
Title:
Date: 11/8/2016
Speaker: Xuanyu Cao
Title: Decentralized Sparse Multitask RLS over Networks
Date: 11/15/2016
Speaker: Xinshan Zhu
Title: Blind Image inpainting Forensics Using Multi-Link Feature
Image inpainting has been a very effective technique for image editing applications. For security considerations, image inpainting forensics becomes more and more important. In this work, we address how to identify the inpainted patch by the extracted feature vector. We use a probability-based classification scheme. In the scheme, the extracted feature signal needs to be statistically characterized and the classification rule is derived based on the stochastic model to be established. Then, in order to effectively distinguish an inpainted image patch from its reference patches, we use a clustering-based post processing operation to remove false alarmed reference patches. Moreover, we present the overall inpainting level (OIL) over the final suspicious clusters to assess how significant the tested image is inpainted. Simulations on real images are conducted to validate the effectiveness of the proposed scheme.
Date: 11/22/2016
Speaker: Xin Liao
Title: Color Image Steganography
In tradition steganographic schemes, a true color image is implicitly thought to be comprised of RGB three channels, and the payload is assigned equally. In fact, the steganographic security of color image relates to not only data-embedding algorithm, but also different channels payload partition. In this paper, a new channel-dependent payload partition model based on minimizing embedding distortion is proposed, so as to assign different embedding capacity among RGB channel instead of average partition. To substantiate our model, we investigate an efficient payload partition strategy, called Channels Distortion Sorting (CDS) strategy, which can be implemented together with state-of-art image steganographic methods. The distortion costs of three channel pixel are ranked to an ascending order, and then the frontal pixels for data embedding are untilized. A theoretical proof is given to ensure the effectiveness of the proposed channel-dependent payload partition. Experimental results show that the steganographic security of the proposed steganographic schemes incorporated with our proposed CDS strategy can outperform the previous ones.
Date: 11/29/2016
Speaker: Yu Yang
Title: Wireless power transfer system modelling and performance evaluation
The urgency of addressing the short usage time of mobile devices and the existence of many market opportunities have motivated both the industry and academia to directly huge efforts toward developing technologies for wireless power transfer (WPT). This research evaluates the power transmission efficiency of the WPT segments which is in support of wideband or ultra-wideband (UWB) system by means of analytically compute the time-averaging output direct current (DC) power at the receiver loads. In order to investigate the impacts of array configurations, propagation channels and rectifier circuits on the transmission efficiency, different array configurations, UWB channels and rectifier circuits are correspondingly modeled. Several examples are implemented and compared by numerical calculations.

### Summer 2016

Date: 06/22/2016
Speaker: Ke Xiong
Title:
Date: 06/29/2016
Speaker: Zhung-Han Wu
Title:
Date: 07/06/2016
Speaker: Yi Han
Title:
Date: 07/13/2016
Speaker: Chen Chen
Title:
Date: 07/20/2016
Speaker: Xuanyu Cao
Title: Iterative Auction Mechanism for Data Trading
In the big data era, it is crucial to distribute the vast amount of data to the heterogeneous users with different demands effectively. To achieve this goal, data owners, collectors and users should cooperate to trade data efficiently. Unfortunately, real world agents are often selfish and seek to maximize their own utilities instead of the overall system performance. As such, an intelligent mechanism is imperative to guild the agents to trade data efficiently. In this work, we present an iterative auction mechanism to elicit the private information of the selfish agents and guild them to trade data in a socially optimal way. The mechanism is theoretically shown to converge to the social optimum. Additionally, the mechanism is demonstrated to possess appealing economic properties including efficiency, individual rationality and weakly budget balance. The mechanism can be further extended to the non-exclusive data trading scenario, where the same data can be distributed to many users. Simulations and real data experiments are conducted to validate the effectiveness and economic properties of the proposed mechanism.
Date: 07/27/2016
Speaker: Qinyi Xu
Title:
Date: 08/03/2016
Speaker: Feng Zhang
Title:
Date: 08/10/2016
Speaker: Xinshan Zhu
Title: Image Inpainting Forgery Forensics: Feature Extraction and Clustering
Image authentication is the main concern in digital multimedia era. Image inpainting has been a very effective technique for image editing applications. For security considerations, image inpainting forensics becomes more and more important. In this work, we address the feature extraction and clustering for inpainting forensics. First, the multiple linking features are presented and their descriptors are defined by combing the normalized cross correlation and the zero connectivity between two patches. The presented feature vector is also weighted by the distance of patches in order to reduce the negative effect from the neighboring patches. Then, the stochastic model for the extracted features is established by using a mixture of mixtures of Gaussian and uniform distributions, and the model parameters are estimated by expectation maximization. With this model, the inpainting probability of a patch is obtained and the inpainting decision is derived while the expectation maximization process converges. Simulations on real images are conducted to validate the effectiveness of the proposed scheme.

### Spring 2016

Date: 02/18/2016
Speaker: Chen Chen
Title:
Date: 02/25/2016
Speaker: Chunxiao Jiang
Title:
Date: 03/10/2016
Speaker: Zhung-Han Wu
Title:
Date: 03/31/2016
Speaker: Hang Ma
Title: Distributed Low Complexity Baseband Signal Compression for Cloud Radio Access Network
The explosive growth of wireless traffic requires revolutionary wireless communication techniques. The cloud radio access network (C-RAN) is a solution to leverage the spatial multiplexing gain by utilizing a number of antennas distributed in a certain area. However, in most of the current literature, due to the limited front-haul capacity, each remote radio head (RRH) can use only a single antenna, and thus the total number of antennas can be utilized are limited. In this work, we propose a distributed compression and symbol detection scheme for the uplink of a multi-antenna C-RAN where each RRH uses multiple antennas. At each RRH, the baseband signals from multiple antennas are compressed into fewer components so that more bits can be allocated to each value, and the quantization noise power caused by the front-haul link capacity deficit is reduced. The compression is low complexity which can be realized using basic buffering and adding, and does not require channel information. Therefore, the low deployment cost feature of C-RAN is preserved. As a result, massive antennas can be utilized by deploying a lot of multi-antenna RRHs, which provide spatial diversity that assists the detection phase which happens in the baseband units (BBUs). In the symbol detection phase, the corresponding LMMSE weight vectors are designed to detect the symbols from the compressed baseband signal. A parallel interference cancellation algorithm is proposed to further improve the accuracy of the symbol detection. Numerical results show that the proposed scheme is efficient in tackling the front-haul capacity challenge. We also apply the proposed scheme to LTE based multi-antenna C-RAN, where we find that the system can utilize larger bandwidth with limited front-haul capacity.
Date: 04/07/2016
Speaker: Ke Xiong
Title: Beamforming design for high-mobility wireless communications
To leverage the benefits of beamforming for high-mobility communication scenarios, we propose a wide beamforming paradigm, which is able to generate a wide beam on the moving reliever, so that a high-mobility user can obtain high received signal-to-noise ratio (SNR) within the broaden coverage region provided by the wide beam. An optimization problem is formulated to minimize the consumed power by jointly optimizing the transmit and the receiving beam vectors while guarantee the required SNR of the moving user for a given time period. Since the problem is with infinite constraints and has no known solution method, we design an efficient algorithm to alternatively optimize the transmit and receiving beam vectors by using semidefinite relaxation with an iterative manner. Simulation results show that, using our proposed wide beamforming scheme, the required SNR of a high-mobility user within a given time period can be guaranteed. Compared with traditional scheme, with the same power consumption, our proposed scheme can enhanced the SNR greatly within the interest region. Besides, by increasing the number of transmit antennas, the consumed power can be decreased, but the decreasing rate decreases with the increment of the the number of transmit antennas.
Date: 04/14/2016
Speaker: Xuanyu Cao
Title: Iterative Auction Mechanism for Data Trading
In the big data era, it is crucial to distribute the vast amount of data to the heterogeneous users with different demands effectively. To achieve this goal, data owners, collectors and users should cooperate to trade data efficiently. Unfortunately, real world agents are often selfish and seek to maximize their own utilities instead of the overall system performance. As such, an intelligent mechanism is imperative to guild the agents to trade data efficiently. In this work, we present an iterative auction mechanism to elicit the private information of the selfish agents and guild them to trade data in a socially optimal way. The mechanism is theoretically shown to converge to the social optimum. Additionally, the mechanism is demonstrated to possess appealing economic properties including efficiency, individual rationality and weakly budget balance. So far, the simulations and real data experiments haven't been conducted yet, which will be done later.
Date: 04/21/2016
Speaker: Yi Han
Title:
Date: 04/28/2016
Speaker: Qinyi Xu
Title:
Date: 05/05/2016
Speaker: Feng Zhang
Title:

### Fall 2015

Date: 09/21/2015
Speaker: Ke Xiong
Title:
Date: 09/28/2015
Speaker: Chunxiao Jiang
Title: Privacy or Utility in Data Collection? A Contract Theoretic Approach
With the growing popularity of data mining, privacy has become an issue of growing importance. Privacy can be seen as a special type of goods, in a sense that it can be traded by the owner for incentives. In this paper, we consider a private data collecting scenario where a data collector buys data from multiple data owners and employs anonymization techniques to protect data owners' privacy. Anonymization causes a decline of data utility, therefore, the data owner can only sell his data at a lower price if his privacy is better protected. Can one pursue higher data utility while maintaining acceptable privacy? How to balance the trade-off between privacy protection and data utility is an important question for big data. Considering that different data owners treat privacy differently, and their privacy preferences are unknown to the collector, we propose a contract theoretic approach for data collector to deal with the tradeoff. By designing an optimal contract, the collector can make rational decisions on how to pay the data owners, and more importantly, how he should protect the owners' privacy. We show that when the collector requires a large amount of data, he should ask data owners who care privacy less to provide as much as possible data. We also find that whenever the collector requires higher utility of data or the data becomes less profitable, the collector should provide a stronger protection of the owners' privacy. Performance of the proposed contract is evaluated by both numerical simulations and real data experiments.
Date: 10/05/2015
Speaker: Zhung-Han Wu
Title:
Date: 10/19/2015
Speaker: Hang Ma
Title:
Date: 10/26/2015
Speaker: Chen Chen
Title:
Date: 11/02/2015
Speaker: Yi Han
Title:
Date: 11/16/2015
Speaker: Qinyi Xu
Title:
Date: 11/30/2015
Speaker: Xuanyu Cao
Title: A game-theoretic modeling of popularity dynamics
With the explosive growth of big data, it is becoming more and more important to capture people's limited attention and interaction. Numerous items (e.g., online memes, videos, papers) are generated everyday, leading to a rich amount of popularity dynamics (e.g., videos' view count dynamics, hashtags' mention count dynamics, and papers' citation dynamics), which record people's interactions with the items. Some of them go viral while most diminish quickly. It is crucial to understand the underlying mechanisms of popularity dynamics. In this work, we propose a game-theoretic model to analyze and understand popularity dynamics. The model takes both the instantaneous incentives and long-term incentives during people's decision making process into account. We theoretically prove that the proposed game possesses a unique symmetric Nash equilibrium (SPE), which can be computed via a backward induction algorithm. We also demonstrate that, in the SPE, the interaction rate first increases and then decreases, which coincides with the observations from real data. Finally, by using simulations as well as experiments based real data (including Twitter hashtag data and paper citation data from Web of Science), we validate the effectiveness of the theory. The theoretical dynamics fit the real data well and can even predict future dynamics data.
Date: 12/07/2015
Speaker: Feng Zhang
Title:

### Summer 2015

Date: 06/10/2015
Speaker: Xiaoyu Chu
Title: Fundamental Limits in Multimedia Forensics and Anti-forensics
As the use of multimedia editing tools increases, people become questioning the authenticity of multimedia content. This is specially a big concern for authorities, such as law enforcement, new reporter and government, who constantly use multimedia evidence to make critical decisions. To verify the authenticity of multimedia content, many forensic techniques have been proposed to identify the processing history of multimedia content under question. However, as new technologies emerge and more complicated scenarios are considered, the limitation of multimedia forensics has been gradually realized by forensic researchers. It is the inevitable trend in multimedia forensics to explore the fundamental limits. In this dissertation, we propose several theoretical frameworks to study the fundamental limits in various forensic problems. Specifically, we begin by developing empirical forensic techniques to deal with the limitation of existing techniques due to the emergence of new technology, compressive sensing. Then, we go one step further to explore the fundamental limit of forensic performance. Two types of forensic problems have been examined. In operation forensics, we propose an information theoretical framework and define forensicability as the maximum information features contain about hypotheses of processing histories. Based on this framework, we have found the maximum number of JPEG compressions one can detect. In order forensics, an information theoretical criterion is proposed to determine when we can and cannot detect the order of manipulation operations that have been applied on multimedia content. Additionally, we have examined the fundamental tradeoffs in multimedia antiforensics, where attacking techniques are developed by forgers to conceal manipulation fingerprints and confuse forensic investigations. In this field, we have defined concealability as the effectiveness of anti-forensics concealing manipulation fingerprints. Then, a tradeoff between concealability, rate and distortion is proposed and characterized for compression anti-forensics, which provides us valuable insights of how forgers may behave under their best strategy.
Date: 06/24/2015
Speaker: Yi Han
Title:
Date: 07/01/2015
Speaker: Zhung-Han Wu
Title:
Date: 07/08/2015
Speaker: Meng-lin Ku
Title:
Date: 07/29/2015
Speaker: Hao Lu
Title: Mode Selection and Resource Allocation for Device-to-Device Communications with Energy Harvesting Constraint
The intra-cell interference due to resources reusing in Device-to-device communications underlaying cellular networks has become research focus nowadays. In our previous works, we established a model to characterize the interference scenarios, based on which we derive the stable throughput region of multiple interactional wireless connections. Then, according to the requirements for energy efficiency in the next generation wireless communications, we introduce energy harvesting process to our model. We still derive the stable throughput region for the D2D communications underlaying cellular networks. The simulation provides an interesting result that the proposed model outperform both TDMA and conventional model in the single D2D pair case. Subsequently, we perform mode selection among D2D pairs to determine the best transmission mode for each D2D pair, when maximizing the spetral efficiency. To solve the mixed-Boolean convex problem, we use Branch-and-Bound method to obtain the optimal solution. For the future work, we are going to search low-complexity sub-optimal solution for mode selection and resource allocation strategy. Simulations should also be done.
Date: 08/05/2015
Speaker: Chen Chen
Title:
Date: 08/12/2015
Speaker: Xuanyu Cao
Title: A Game Theoretic Modeling of Citations
Citation is the most important metric for quantifying scientific impact of papers. In this work, we investigate the mechanism of generating the citation dynamics of individual papers from a game-theoretic perspective. A sequential game theory model is proposed to model the citation process. We prove that the proposed citation game possesses a unique symmetric Nash equilibrium, which can be computed by using a backward induction algorithm. We further show that, under several reasonable assumptions, the citation rate first increases and then decreases, in accordance with the observation from real data.
Date: 08/19/2015
Speaker: Qinyi Xu
Title:
Date: 08/26/2015
Speaker: Feng Zhang
Title:

### Spring 2015

Date: 04/27/2015
Speaker: Wei Li
Title: Two-Way Multi-Relay Networks with stochastic energy harvesting
A optimal transmission policy for two-way multi-relay networks with space-time network coding is investigated by using a stochastic energy harvesting (EH) model. Considering the channel and finite battery conditions, we propose the optimal relay transmission power to minimize the long-term pairwise error probability of the network. The design problem is formulated as a Markov decision process (MDP), and the well-known value iteration approach is used to find the optimal policy. Based on the optimal transmission policy, we analyze the special structures of the optimal relay transmission policy and the diversity order. Simulation results demonstrate that our proposed optimal transmission policy outperforms other naive policies.
Date: 04/13/2015
Speaker: Xiaoyu Chu
Title: Can Orders of Operations be Distinguished? An Information Theoretical Criterion
Multimedia forensics deals with problems of authenticating multimedia content by examining the fingerprints of manipulations. While most existing techniques focus on detecting single manipulation operations, it is often the case that multiple operations have been done during the manipulation. In order to obtain a complete processing history, for these cases, we need to detect not only what operations have been applied, but also the order of when were they applied. However, given the possible modification of latter applied operations on earlier applied ones, it is not always the case that the order can be detected. Therefore, finding when we can and cannot distinguish the order of these operations is an interesting problem. In this work, I'll propose an information theoretical framework and give a mutual information criterion to determine when we can and cannot detect the orders. Three case studies will be examined based on this framework to demonstrate its effectiveness.
Date: 04/06/2015
Speaker: Xuanyu Cao
Title: Community detection in noiseless and noisy graphs: a game-theoretic framework
Many real-world networks have the so-called community structure: groups of nodes form functional modules, e.g., groups of people in social networks, functional modules of cells in biological networks. It is crucial to identify the communities of the networks. Additionally, due to techonogocial constraints, many real networks are noisy: there can be lots of missing links or fake linkes. In this work, we present a game-theoretic treatment of the community detection problem in both noiseless and noisy graphs, where nodes' utilities depend on their community affiliations. The equilibria of the formulated games are derived, which lead to a general probabilistic generative model of the formation of the networks. Given the observation of the networks, we infer the community affiliation of each node using an expectation-maximization algorithm. Simulations and real-world dataset (facebook and DBLP) experiments show that the proposed noise-aware algorithms outperform the state-of-the-art community detection methods on noisy networks.
Date: 03/30/2015
Speaker: Yichen Wang
Title: Statistical delay QoS protection for primary users in cooperative cognitive radio networks
In cooperative cognitive radio networks, where secondary users (SU) can relay the signal of primary user (PU) in exchange for PU’s licensed spectrum, most existing works employed deterministic quality-of-service (QoS) guarantee in terms of minimum required transmission rate for PU’s protection. In this letter, we adopt the statistical QoS requirement characterized by queue-length bound violation probability for PU’s delay QoS provisioning. By applying the effective capacity theory, we further convert PU’s queue-length bound violation probability requirement to the equivalent effective capacity constraint and formulate the corresponding SU’s throughput maximization problem. Then, we obtain the optimal joint time-slot allocation scheme, where the timeslot division adapts to both channel conditions and PU’s delay QoS requirements, such that not only PU’s statistical delay QoS requirement can be well guaranteed, but also SU’s throughput can be optimized. Moreover, we also develop a fixed time-slot allocation scheme which only varies with PU’s delay QoS requirements. Simulation results show that both the optimal and fixed allocation schemes can flexibly perform time-slot divisions according to the delay QoS requirements of PU’s traffic, but the developed optimal scheme outperforms the fixed time-slot allocation scheme.
Date: 03/23/2015
Speaker: Qinyi Xu
Title:
Date: 03/09/2015
Speaker: Chen Chen
Title:
Date: 03/02/2015
Speaker: Hang Ma
Title:
Date: 02/16/2015
Speaker: Yi Han
Title:
Date: 02/09/2015
Speaker: Zhung-Han Wu
Title:
Date: 02/02/2015
Speaker: Hao Lu
Title: Admission Control of Device-to-device Communications underlaying Cellular Networks
The intra-cell interference due to resources reusing in Device-to-device communications underlaying cellular networks has become research focus nowadays. Existing works are based on information-theoretic studies under full rate transmission of both D2D pairs and cellular users and those resources allocation schemes are performed statically at base station. In our work, we propose a generalized two-connection underlaying model via stable throughput analysis, accounting packet arrival process in practical networks. Furthermore, we propose a dynamic admission control and power allocation scheme for D2D connections underlaying cellular networks. Subsequently, the two-connection model is extended to multiple D2D pairs scenarios. Simulation results will show that the access probabilities of our proposed scheme significantly increase when comparing with conventional admission criterion.

### Fall 2014

Date: 12/11/2014
Speaker: Yanxiang Jiang
Title: Energy efficient resource allocation in uplink Massive MIMO system
In this work, resource allocation for uplink massive MIMO communications systems is investigated. By adopting MMSE receiver at BS, SINR performance under perfect and imperfect CSI in both instantaneous and asymptotic case are analyzed, respectively. For perfect CSI case, certain approaches are considered to tackle the problem of variable coupling in resource allocation. Besides, spectral fairness among users is also considered during the energy efficiency optimization process. For imperfect CSI case, the effect of pilot power on EE is considered.
Date: 11/13/2014
Speaker: Qinyi Xu
Title:
Date: 11/06/2014
Speaker: Chen Chen
Title:
Date: 12/11/2014
Speaker: Wei Li
Title: On the outage probability for two-way relay networks with stochastic energy harvesting
A optimal transmission policy for two-way relay networks is investigated by using a stochastic energy harvesting (EH) model. Considering the channel and finite battery conditions, we propose the optimal relay transmission power to minimize the long-term outage probability of the network. The design problem is formulated as a Markov decision process (MDP), and the well-known value iteration approach is used to find the optimal policy. Based on the optimal transmission policy, we analyze the special structures of the optimal relay transmission strategy and the outage probability performance. Simulation results demonstrate that our proposed optimal transmission policy outperforms other naive policies.
Date: 10/23/2014
Speaker: Xiaoyu Chu
Title: Detect Order of Blurring and Resizing
Multimedia forensics deals with problems of authenticating multimedia content by examining the fingerprints of manipulations. While most existing techniques focus on detecting single manipulation operations, it is often the case that multiple operations have been done during the manipulation. Therefore, estimating the processing history of a multimedia file when multiple operations have been applied on is an important problem. Moreover, detecting the order of these operations is the key to identify the processing history. In this talk, I'll propose a framework for detecting the order of operations, wherein conditional fingerprints will be defined, as well as the fingerprints in single operation cases. Then, the case of detecting the order of blurring and resizing will be examined. I will revisit the identification table proposed in my last presentation to determine if this order can be detected. After that, specific detectors will be designed to identify the order of blurring and resizing. The last but not the least, channel based model will be proposed to obtain the optimal parameters for the detector which yields the highest mutual information or capacity.
Date: 10/16/2014
Speaker: Xuanyu Cao
Title: Quantifying Scientific Impact with Short-term Data
Citation is perhaps the mostly used metric to evaluate the scientific impact of papers. Various measures of the scientific impact of researchers and journals rely heavily on the citations of the papers. Furthermore, in many practical applications, people may need to know not only the current citations of a paper, but also an estimate of its future citations. However, the complex heterogeneous temporal patterns of the citation dynamics make the predictions of future citations rather difficult. In this paper, we present a simple data-driven non-parametric method to predict future citations of individual papers. With short-term (e.g. 3 years after the paper is published) citation data, our approach can already give accurate estimate of future citations, outperforming state-of-the-art prediction methods significantly. Extensive experiments confirm the robustness of our approach across various journals of different disciplines. Our method can be applied to policy making to foresee the future scientific impact of researchers.
Date: 10/02/2014
Speaker: Yi Han
Title:
Date: 09/25/2014
Speaker: Hang Ma
Title:

### Summer 2014

Date: 08/20/2014
Speaker: Qinyi Xu
Date: 08/13/2014
Speaker: Chen Chen
Date: 08/06/2014
Speaker: Yanxiang Jiang
Title: Energy Efficient Resource Optimization in Small Cell System
In this work, a distributed energy-efficiency power allocation in small cell networks with a minimum required data rate is studied. A non-cooperate game is modeled to describe the system model. The considered non-convex program for each small base station (SBS) is transformed into a convex program by utilizing the parameter-free fractional programming which uses the concept of perspective function. Then, the constrained convex optimization problem is converted into unconstrained convex optimization problem by exploiting mixed penalty function method. Furthermore, the unconstrained optimization problem is solved by Newton method. Besides, the existence and the uniqueness of the nash equilibrium and the convergence of Newton method are derived. Simulation results illustrate that the proposed power allocation algorithm not only converges to the nash equilibrium, but reaches nearly the maximum system energy efficiency.
Date: 07/30/2014
Speaker: Xiaoyu Chu
Title: Fundamental Limits of Order Forensics
Multimedia forensics seek for fingerprints of manipulations to verify the authenticity of multimedia files. While most forensic techniques are based on binary hypothesis test and can only detect single manipulation operations, many manipulations in reality are done through chains of multiple operations. For a processing chain of multiple operations, the order of these operations is an essential part of determining the complete processing history. Moreover, given that different operations may be applied by different people, determining the order of operations can also provide us information about when and who manipulated the multimedia content. However, difficulties of such problem comes from the effect of one operation on the fingerprints of another operation in the processing chain. Sometimes, the fingerprints of an operation can be entirely removed by another operation applied after it, in which cases it is hard to identify the earlier applied operation. Therefore, when we can or cannot determine the order of operations in a processing chain is an important yet challenging problem. Matthew Stamm had given a formulation of this problem before and proposed some forensic techniques to detect the order of 2 operations. However, his formulation is not suitable to understand the fundamental limits nor obtain a systematic way to determine whether the order can/cannot be detected. Hence, in this work, we re-formulate the problem of order forensics in a more rigorous and scalable way and answer the following questions: 1) How do we determine whether we can or cannot detect the order of operations in a processing chain? 2) What is the best detection performance we can get in the order of operation forensics?
Date: 07/23/2014
Speaker: Hang Ma
Date: 07/16/2014
Speaker: Yi Han
Date: 07/09/2014
Speaker: Wei Li
Title: Optimal Transmission Policy for Two-Way Relay Networks with Stochastic Energy Harvesting
An optimal transmission policy for two-way relay networks is investigated by using a stochastic energy harvesting (EH) model. Considering the channel and finite battery conditions, we propose an optimal relay transmission policy to maximize the long-term achievable sum rate of the network. The design problem is formulated as a Markov decision process (MDP), and the well-known value iteration approach is used to find the optimal policy. Based on the optimal transmission policy, we analyze the expected achievable sum rate and point out a spreading structure for the optimal relay power with respect to the solar panel size. Simulation results demonstrate that our proposed optimal transmission policy outperforms other policies.
Date: 07/02/2014
Speaker: Xuanyu Cao
Title: Evolutionary Information Diffusion over Heterogeneous Social Networks
A huge amount of information, created and forwarded by various people, is propagating through the online social networks every day. It is important to understand the mechanisms of the general information diffusion process over the social networks composed of heterogeneous individuals with different interests, habits and influences. Different from most of the existing works, we investigate the information diffusion from an evolutionary game perspective and try to reveal the underlying principles dominating the complex information diffusion process over the heterogeneous social networks. Modeling the interactions among the heterogeneous users as a graphical evolutionary game, we derive the evolutionary dynamics and the evolutionarily stable states (ESSs) of the diffusion. The different payoffs of the heterogeneous users lead to different diffusion dynamics and ESSs among them, characterizing the heterogeneity observed in real-world datasets. Simulations are conducted to verify our theoretical results. Finally, we test our theory on Twitter hashtag dataset. We observe that our derived evolutionary dynamics fit the data well and can predict the future diffusion data after the peak of the dynamics. The prediction outperforms our previous work on homogeneous modeling, indicating the significance of our heterogeneous model in analyzing the real-world heterogeneous social networks.
Date: 06/25/2014
Speaker: Zhung-Han Wu
Date: 06/13/2014
Speaker: Yang Gao
Title: ON THE DESIGN AND ANALYSIS OF INCENTIVE MECHANISMS IN NETWORK SCIENCE

### Spring 2014

Date: 04/16/2014
Speaker: Yanxiang Jiang
Title: Energy efficient joint resource allocation and power control for D2D communications
In this work, joint resource allocation and power control for energy efficient device-to-device (D2D) communications underlaying cellular networks are investigated. The considered problem is modeled as a nonconvex optimization problem which takes the circuit power consumption and the interference between cellular users and D2D users into consideration. The resource and power are optimized for maximization of the energy efficiency (EE) of D2D communications. By exploiting the properties of fractional programming, the resulting nonconvex optimization problem in fractional form is transformed into an equivalent optimization problem in subtractive form, which leads to an efficient iterative resource allocation and power control scheme. In each iteration, part of the constraints of the EE optimization problem is removed by exploiting a penalty function approach. Then, we propose to transform the equivalent optimization problem into a two-layer optimization problem. In the first layer, the optimal power values are obtained by solving a series of maximization problems through root-finding with or without considering the loss of cellular users' rates. In the second layer, a heuristic resource allocation algorithm is developed to maximize the EE and avoid conflict among the D2D pairs. Simulation results illustrate that the proposed iterative resource allocation and power control scheme converges fast in a small number of iterations and demonstrate the remarkable improvements in term of EE.
Date: 04/02/2014
Speaker: Zhung-Han Wu
Date: 03/26/2014
Speaker: Yang Gao
Title: Understanding the Interaction between Question Answering and Voting in Social Computing: A Case Study of Stack Overflow
Social computing where users generate online content and create values in exchange for virtual or monetary rewards has become ubiquitous on the web. The success of social computing applications largely depends on active user participation, which often takes two forms: producing a piece of content directly (question answering) and voting on existing content. These two forms of user activities interact closely with each other and should be studied jointly. In this talk, we present an extensive form game model that captures the decision making of sequentially arrived users who choose endogenously whether to participate or not and between question answering and voting if participate. We prove theoretically that there exists a unique symmetric subgame perfect equilibrium (SSPE), which is a pure strategy action rule and has a threshold structure in every state of the game. To justify our model, we collect data from Stack Overflow, the most popular focused Q&A site, and show that the predictions of our model are consistent with observations from the data. Finally, we conduct simulations and abstract four design principles that could potentially aid in the design of incentive mechanisms for social computing applications in practice.
Date: 03/12/2014
Speaker: Hang Ma
Date: 03/05/2014
Speaker: Xiaoyu Chu
Title: Fundamental Limits in Multimedia Forensics and Anti-forensics
As there are more and more forensics and anti-forensic techniques been proposed in recent years, the question of "what is our limit" becomes popular. In this proposal, we explore the fundamental limits of multimedia forensics and anti-forensics by building theoretical frameworks. In our preliminary works, we defined forensicability for forensics and concealability for anti-forensics to measure their capabilities. Based on those, specific examples were analyzed to demonstrate the effectiveness of our model, and from which interesting results were found. In our future works, we want to examine the fundamental limits in order forensics and study the interaction between forensicability and concealability.
Date: 02/26/2014
Speaker: Wei Li
Title: Orthogonal Projection Based MIMO Multi-Way Relaying with Relay Antenna Subset Selection
An asymmetric multi-way relay communication scheme using orthogonal projection and analog network coding is proposed. The scheme is designed for the wireless sensor network, where one sink node exchanges information with $N$ user nodes via a relay station and neither the sink node nor user nodes need complex transceivers. We show that the minimal number of the relay antennas, $M_{min}$, is only $2N-1$, which is smaller than that of the total data streams. Moreover, when the number of the relay antennas, $M$, is larger than the minimum, we develop a paired throughput based Max-Min criterion with low computational complexity to select the optimal relay antenna subset and guarantee the full diversity gain of $M-M_{min}+1$. The numerical results demonstrate the high outage performance and throughput of our proposed communication scheme and antenna subset selection criterion.
Date: 02/19/2014
Speaker: Xuanyu Cao
Title: Cognitive Radio Networks with Heterogeneous Secondary Users: How to Procure and Price the Spectrum?
In this paper, we investigate the optimal spectrum procurement and pricing from the perspective of a cognitive mobile virtual network operator (C-MVNO), which is a second market between the spectrum owner and the secondary users (SUs). The spectrum procurement consists of spectrum leasing and spectrum sensing, the latter of which has uncertain outcome. The SUs are assumed to be heterogeneous in their valuations and demands of the spectrum, which is generally the case in reality. Hence, we use differentiated pricing among the heterogeneous SUs to improve the profit of the C-MVNO and allow the C-MVNO to perform necessary admission control. Modeling the spectrum procurement and trading procedure as a five-stage Stackelberg game, we analyze the optimal decisions for the C-MVNO by using backward induction. The optimal decisions of spectrum sensing, spectrum leasing, admission control and differentiated pricing are derived and an algorithm is proposed to compute those optimal decisions efficiently. Our theoretical results are also corroborated by numerical experiments and a threshold structure of the solution is observed.
Date: 02/05/2014
Speaker: Yi Han

### Fall 2013

Date: 11/26/2013
Speaker: Zhung-Han Wu
Title: Deep Neural Networks: Application on Image Forensics
Information forensics uses the fingerprints left behind by manipulations to identify the authenticity of the information. Finding an effective fingerprint is not an easy task and is time consuming. This leads to a slowdown on the discovery of new forensic fingerprints. Recent deep learning technology has shown promising results in both representation learning and classification scenarios. More importantly, the technology can discover the underlying structures in information which is helpful in information forensics. In this talk, a brief introduction to deep learning would be presented and the possible application to information forensics would be proposed.
Date: 11/19/2013
Speaker: Chanho Yoon
Title: Simple cooperative filer-and-forward beamforming for SCBT systems
In recent years, single carrier block transmission (SCBT) has become an interesting and complementary alternative to OFDM for combating frequency selective fading channels. In uncoded case, using non-linear equalizer techniques at the receiver side, error rate performance can even surpass that of OFDM. However, the result becomes reversed in channel coded case despite various linear/non linear equalization methods devised for pushing the capacity to its limit in frequency selective fading. The performance of SCBT is still far from its achievable matched filter bound. Therefore, we suggest cooperative filer-and-forward beamforming scheme to overcome the frequency selective fading channel, and also derive a simple pre-filter used at the destination node.
Date: 11/12/2013
Speaker: Chunxiao Jiang
Title: Evolutionary Information Diffusion over Social Networks
Social networks have become ubiquitous in our daily life, as such it has attracted great research interests recently. A key challenge is that it is of extremely large-scale with tremendous information flow, creating the phenomenon of "Big Data". Under such a circumstance, understanding information diffusion over social networks has become an important research issue. Most of the existing works on information diffusion analysis are based on either network structure modeling or empirical approach with dataset mining. However, the information diffusion is also heavily influenced by network users' decisions, actions and their socio-economic connections, which is generally ignored in existing works. In this work, we propose an evolutionary game theoretic framework to model the dynamic information diffusion process in social networks. Specifically, we analyze the framework in uniform degree and non-uniform degree networks and derive the closed-form expressions of the evolutionary stable network states. Moreover, the information diffusion over two special networks, Erdos-Renyi random network and the Barabasi-Albert scale-free network, are also highlighted. To verify our theoretical analysis, we conduct experiments by using both synthetic networks and real-world Facebook network, as well as real-world information spreading dataset of Memetracker. Experiments shows that the proposed game theoretic framework is effective and practical in modeling the social network users' information forwarding behaviors.
Date: 11/05/2013
Speaker: Hang Ma
Date: 10/08/2013
Speaker: Yang Gao
Title: To Answer or to Vote? A Game-Theoretic Analysis of Sequential Activities in Social Computing
Social computing where users generate online content and create values in exchange for virtual or monetary rewards has become ubiquitous on the web. It is thus of great importance to model and analyze user behaviors in social computing systems. A typical scenario involves sequentially arrived users who choose either to produce a piece of content (to answer) or to vote on existing content or none of the two. In this talk, we model such sequential decision making as an extensive form game with infinite horizon. We prove theoretically that there exists a unique symmetric subgame perfect equilibrium (SSPE), which is a pure strategy action rule and has a threshold structure in every state of the game. We further develop a dynamic programming algorithm that obtains the unique SSPE with linear complexity.
Date: 10/01/2013
Speaker: Xiaoyu Chu
Title: Information Theoretical Analysis of Fundamental Limit in JPEG Compression Forensics
As forensic field has been filled with all kinds of detecting and identifying techniques to verify the authenticity of multimedia contents, researchers have begun wondering the forensicability of investigators, i.e., the limit of their capability, if any. In this work, we introduce an information theoretical model to the forensic system and use mutual information to quantify forensicability. Specifically, we present our model by considering the example of detecting multiple JPEG compressions. In this case, we derived the mutual information and the relationship between conditional entropy and error rate. We further defined the expected perfect detection to relate theoretical error rate and experimental performance. Our simulation results show that mutual information depends on the quantization step sizes used in each compression. We particularly found the pattern of those quantization step sizes yielding the lowest and highest mutual information. More importantly, we showed that when the number of compressions we want to detect is larger than 30 (actually somewhere between 20 and 30), it is sure that NO perfect detection (in expected sense as we defined) can be achieved.
Date: 09/24/2013
Speaker: Yi Han
Date: 09/10/2013
Speaker: Yu-Han Yang
Title: Joint Waveform Design and Interference Pre-Cancellation for Time-Reversal Systems
In wideband communication systems, the time-reversal (TR) technique can boost the signal-to-noise ratio (SNR) at the receiver with simple single-tap detection. It has been shown that optimal waveform design can significantly improve the system performance of TR systems. However, when the symbol rate is very high, the severe inter-symbol interference (ISI) still limits the performance at high power region. In this work, we study the joint waveform design and interference pre-cancellation by exploiting the symbol information to further improve the performance. In the proposed joint design, the causal ISI is subtracted by interference pre-cancellation and the anti-causal ISI can be further suppressed by the waveform design with the more abundant degrees of freedom. The transmitter utilizes the information of previous symbols to enhance the signal quality while the receiver structure remains simple. In the multi-user scenario, both the inter-user interference (IUI) and ISI can be similarly categorized by its causality, and then be tackled accordingly by the proposed joint design. The resulting multi-user waveform design is a non-convex optimization problem, for which two iterative algorithms are proposed and both are guaranteed to converge to suboptimal solutions. Simulation results validate the convergence behavior and demonstrate the remarkable performance improvement over the non-joint waveform design in the previous work.

### Summer 2013

Date: 8/14/2013
Speaker: Tong Zhou
Title: Time Mute in Closed Access Femto and Macro Co-existent Networks
In this work, we present an analytical framework to evaluate the coverage probability in closed access femto and macro co-existent networks. We analyze altruistic interference mitigation from femto tier, and derive the optimal operating parameters for downlink power control and time mute schemes. Then we formulate the two-tier interference coordination as price games, and obtain the close form of the Nash equilibriums. Simulation results verify the derivations and show that time mute scheme performs better than power control scheme in dealing with the indoor coverage-hole problem.
Date: 7/31/2013
Speaker: Zhung-Han Wu
Title: Deep Neural Networks and Information Forensics
Information forensics uses the fingerprints left behind by manipulations to identify the authenticity of the information. Finding an effective fingerprint is not an easy task and is time consuming. This leads to a slowdown on the discovery of new forensic fingerprints. Recent deep learning technology has shown promising results in both representation learning and classification scenarios. More importantly, the technology can discover the underlying structures in information which is helpful in information forensics. In this talk, a brief introduction to deep learning would be presented and the possible application to information forensics would be proposed.
Date: 7/24/2013
Speaker: Sha Wei
Title: Power Scheduling for Energy Harvesting Systems with Battery Capacity Constraints
Energy harvesting (EH) technology, which obtains energy from surrounding environment, becomes an effective method to provide almost unlimited energy supply and extend the lifetime of wireless sensor networks (WSNs). Comparing to conventional energy-limited WSNs, the sensor node equipped with energy harvester should satisfy two new transmission power constraints, namely the EH causality and battery capacity constraints. In this work, we study the power scheduling problem to optimize the outage probability in both battery-unlimited and battery-limited scenarios under energy harvesting constraints. In particular, we first model the average outage probability minimization problem over finite horizon, and then reformulate the problem by minimizing its upper bound, which converts the non-convex problem to a more efficient convex optimization problem. For infinite battery scenario, we obtain the necessary optimality conditions of optimal power management solution through Lagrange method. Based on the slotwise structure of optimal transmission power and slot boundary identifying, we propose a slotwise equally power scheduling result to minimize the upper bound of outage probability in the battery-unlimited scenario. In addition, for a more general finite battery scenario, we study the slot boundary through sequentially minimal optimization method, and present a slotwise bottleneck power scheduling strategy to minimize the upper bound of outage probability. Simulation results show that, our power allocation schemes achieve better outage performance compared to other alternative strategies, e.g., best-effort, fixed-saving and random saving strategies.
Date: 7/03/2013
Speaker: Hang Ma
Title: Time-Reversal Based Small Cells and Distributed Antenna System
Current cellular systems suffer from the problems of poor indoor coverage and insufficient capacity for the massive wireless traffic. Small cells and distributed antenna systems are potential solutions for these problems by bringing antennas closer to the users. However, due to the broadcasting nature of wireless communication, smart coordination between these antennas is needed in order to suppress the mutual interferences. Due to the spatial and temporal focusing effect of time-reversal communication, it is an ideal candidate for the air interface of these systems. In this work, the time-reversal technology is applied to the small cells and the distributed antenna system, which alleviates much complexity in implementation. Moreover, numerical results show that the time-reversal based small cells and distributed antenna system achieve satisfactory performance.
Date: 6/25/2013
Speaker: Feng Han
Title: Energy-Efficiency Optimization in Green Wireless Communications
The rising energy concern and the ubiquity of energy-consuming wireless applications has sparked a keen interest in the development and deployment of energy-efficient eco-friendly wireless communication technology. Green Wireless Communications aims to find innovative solutions to improve energy efficiency, and to relieve/reduce the carbon footprint of wireless industry, while maintaining/improving performance metrics. In this dissertation research, towards the evolution of the Green Wireless communications, we study a very promising green communications paradigm, the time reversal system. We first propose a concept of time reversal division multiple access (TRDMA) as a novel wireless media access scheme for broadband networks, and investigate its fundamental theoretical limits. Motivated by the great potential of the TRDMA, we develop a unique asymmetric architecture for the TRDMA based multiuser networks. A channel probing method is presented as a hand-shaking process between the Base Station and terminal users. The unique asymmetric architecture shifts the most complexity to the BS at both downlink and uplink schemes, facilitating very low-cost terminal users in the networks. To further enhance the system performance, a 2D parallel interference cancellation scheme is invented to mitigate both the inter-symbol interference and inter-user interference for the TRDMA uplink at the Base Station. Both theoretical analysis and simulation results demonstrate that the proposed interference cancellation scheme can explore the inherent structure of the interference signals, and therefore efficiently improve the resulting SINR and system performance. Lastly, we explore the energy-saving potential for the cellular infrastructure by incorporating the cooperative BS operation and dynamic BS switching. Four different switching patterns are proposed and studied in the context of cooperative communications. It is shown that significant energy saving potential can be achieved by using BS cooperation and dynamic BS switching without sacrificing the quality of service of the users.
Date: 6/19/2013
Speaker: Chunxiao Jiang
Title: Indian Buffet Game with Negative Network Externality and Non-Bayesian Social Learning
How users in a dynamic system perform learning and make decision become more and more important in numerous research fields. Although there are some works in the social learning literatures regarding how to construct belief on an uncertain system state, few study has been conducted on incorporating social learning with decision making. Moreover, users may have multiple concurrent decisions on different objects/resources and their decisions usually negatively influence each other's utility, which makes the problem even more challenging. In this paper, we propose an Indian Buffet Game to study how users in a dynamic system learn the uncertain system state and make multiple concurrent decisions by not only considering the current myopic utility, but also taking into account the influence of subsequent users' decisions. We analyze the proposed Indian Buffet Game under two different scenarios: one on customers requesting multiple dishes without budget constraint and the other with budget constraint. For both cases, we design recursive best response algorithms to find the subgame perfect Nash equilibrium for customers and characterize special properties of the Nash equilibrium profile under homogeneous setting. Moreover, we introduce a non-Bayesian social learning algorithm for customers to learn the system state, and theoretically prove its convergence. Finally, we conduct simulations to validate the effectiveness and efficiency of the proposed algorithms.

### Spring 2013

Date: 5/1/2013
Speaker: Xiaoyu Chu
Title: Exploring the Fundamental Limit of Multimedia Forensics
Multimedia Forensics is an field where researchers try to extract forensic information from a suspicious multimedia signal to reveal its processing history. Whenever a multimedia file goes through some processing, a certain trace will be embedded in the signal inadvertently. The forensic analysts, on the other hand, extract different features from a given multimedia signal to obtain forensic information. However, as the number of possible processings increases, the multimedia signal, which acts as the information carrier, must have some fundamental limit on how much it can carry to the forensic analysts so that the processing history can be revealed correctly. In this work, we explore such fundamental limit by modeling the multimedia forensic system as a communication system and use channel capacity to quantify the amount of forensic information one can extract from a certain feature. Specifically, we will use image compression as an example of this communication and study the fundamental limit on detecting multiple JPEG compressions. In this talk, I will mainly focusing on explaining how do we modeling the multimedia forensic communication system. Some preliminary results will be shown in the end, together with possible future directions.
Date: 4/24/2013
Speaker: Tong Zhou
Title: Time-Muting Game in CSG Femto-Macro Co-existence Networks
In this work, we study the inter-cell interference coordination scheme to deal with the coverage hole problem in the in CSG Femto-Macro co-existence networks. We derive the coverage probability of indoor Macro UE on the shared channel, and that of Femto UE on shared/dedicated channel.The interference coordination is formulated as the power game and time-muting game respectively. The Nash Equilibrium of the power game are derived.
Date: 4/17/2013
Speaker: Sha Wei
Title: Outage Probability Optimization for a P2P Channel under Energy Harvesting Constraints
Energy harvesting (EH) technology, which obtains energy from surrounding environment, becomes an effective method to provide almost unlimited energy supply and extend the lifetime of wireless sensor networks (WSNs). Comparing to conventional energy-limited WSNs, the sensor node equipped with energy harvester should satisfy two new transmission power constraints, namely the EH causality and battery capacity constraints. In this work, we study the power allocation problem to optimize the outage probability in both battery-unlimited and battery-limited scenarios under energy harvesting constraints. And in each scenario, we discuss both the equal and unequal transmission rates cases. In particular, we first model the average outage probability minimization problem over finite horizon, and then reformulate the problem by minimizing its upper bound, which converts the non-convex problem to a more efficient convex optimization problem. Next, we show that the optimal power allocation solution is related to the global minimum of an energy-and-time function which is defined as arithmetic average of harvested energy from current time slot to the arbitrary time slot in the future before the end of transmission. Based on such properties, we propose a Piecewise Equally power allocation strategy to minimize the upper bound of outage probability in the battery-unlimited scenario. In addition, for a more general battery-limited scenario, we present a Piecewise Bottleneck power allocation strategy. The aforementioned strategies are extended to unequal transmission rates cases by substituting the real harvested energy with the defined normalized energy. Simulation results show that, the proposed power allocation schemes achieve better outage performance compared to other alternative strategies, e.g., best-effort, fixed-saving and random saving strategies.
Date: 4/10/2013
Speaker: Chunxiao Jiang
Title: Optimal Pricing Strategy for Operators in Cognitive Femtocell Networks
Cognitive femtocell has been envisioned as a promising technology for covering indoor environment and assisting heavy loaded macrocell network. Although lots of technical issues of cognitive femtocell network have been studied, e.g., spectrum sharing, interference mitigation, etc, the economic issues that are very important for practical femtocell deployment have not been well investigated in the literatures. In this paper, we focus on the pricing issues in cognitive femtocell network, and propose a two-tier pricing game theoretic framework with two models: static pricing model and dynamic pricing model. In static pricing model, we derive the close-form expressions for pricing and demand functions, as well as the Nash equilibrium pricing strategies for both macrocell and femtocell operators. In dynamic pricing model, we first model the cognitive users' network access behavior as a 2-dimensional Markov decision process and propose a modified value iteration algorithm to find the best strategy profiles for cognitive users. Based on the analysis of users' behavior, we further design an iterative gradient descent algorithm to find the Nash equilibrium pricing strategies for both macrocell and femtocell operators. Simulation results verify our theoretic analysis and show that the proposed algorithm in dynamic pricing model can quickly converge to the Nash equilibrium prices.
Date: 3/27/2013
Speaker: Wei Guan
Title: Two-Way Relay Channel With Network-Coded ARQ
Random channel fading may cause serious channel outage, and automatic repeat-request (ARQ) mechanism is widely used to mitigate packet loss. For two-way relay channel (TWRC), ARQ can be initiated either by the source nodes or by the relay nodes, and it is usually subject to certain maximum transmission constraint. In this work, we study the throughput of TWRC with different ARQ strategies. Beyond orthogonal relaying, we investigate two network-coded ARQ schemes, where the packets intended for different end terminals are combined at the relay nodes whenever possible. For single-relay networks, we derive the closed-from throughput when ARQ is subject to per-hop or end-to-end maximum transmission constraint. We demonstrate that network coding can greatly improve the system throughput, but the throughput gain is upper bounded. Besides, we come up with a near-optimum power allocation scheme to maximize the throughput. For multi-relay networks, we show that successive relaying strategy suffers great throughput loss when the frame length is much smaller than the number of relays, and we develop a hybrid network coding scheme to fully leverage the network coding gain.
Date: 3/13/2013
Speaker: Feng Han
Title: A Multiuser TRDMA Uplink System with 2-Dimensional Parallel Interference Cancellation
The concept of Time Reversal Division Multiple Access (TRDMA) has recently been proposed as a promising medium access technology for the multi-user wireless broadband communications. Compared with the existing multi-carrier technology like OFDM/OFDMA, the TRDMA provides a cost-effective single-carrier alternative technology to combat the inter-symbol interference (ISI) for broadband communications, and at the same time leverages the degrees of freedom in a large number of multi-path to form a unique high-resolution spatial focusing effect. Previous work on TRDMA mainly focus on the multi-user downlink system. In this work, we first introduce a TRDMA-based multi-user uplink architecture and then propose a 2-dimensional (2D) parallel interference cancellation scheme to further enhance the system performance. The TRDMA uplink architecture keeps the cost of end-users at a minimum level, and reuses the processing power at the base station (BS) that has already been made available for the downlink. The proposed 2D parallel interference cancellation scheme utilizes the tentative decisions of detected symbols to effectively cancel the interference in both the time dimension (ISI) and the user dimension (inter-user interference (IUI)), which significantly improve the bit-error-rate performance in the high signal-to-noise-ratio (SNR) regime. To further improve the BER performance, a multi-stage processing can be implemented by cascading multiple stages of the proposed 2D interference cancellation, with a total delay that increases linearly with the number of stages, but independently with the number of users. The BER performance of the single-stage cancellation is analyzed, and the approximated theoretical result well matches the simulation results. Simulation results are provided for up-to 3 stages of interference cancellation and compared with the basic TRDMA system without interference cancellation.
Date: 2/27/2013
Speaker: Matt Stamm
Title: Digital Information Forensics and Anti-Forensics
As the use of digital multimedia content such as images and video has increased, so has the means and the incentive to create digital forgeries. Presently, powerful editing software allows forgers to create perceptually convincing digital forgeries. Accordingly, there is a great need for techniques capable of authenticating digital multimedia content. In response to this, researchers have begun developing digital forensic techniques capable of identifying digital forgeries. These forensic techniques operate by detecting imperceptible traces left by editing operations in digital multimedia content. In this talk, I will discuss several new digital forensic techniques to detect evidence of editing in digital multimedia content. I will begin by identifying the fingerprints left by pixel value mappings and show how these can be used to detect the use of contrast enhancement in images. These fingerprints can be used to perform a number of additional forensic tasks such as identifying cut-and-paste forgeries, detecting the addition of noise to previously JPEG compressed images, and estimating the contrast enhancement mapping used to alter an image. Additionally, I will examine the problem of multimedia security from the forger’s point of view. I will demonstrate that an intelligent forger can design anti-forensic operations to hide editing fingerprints and fool forensic techniques. To demonstrate this, I will discuss my anti-forensic technique to remove compression fingerprints from digital images and show that this technique can be used to fool several state-of-the-art forensic algorithms. I will also examine the problem of detecting frame deletion in digital video and develop both a technique to detect frame deletion and an anti-forensic technique to hide frame deletion fingerprints. I will show that this anti-forensic operation leaves behind fingerprints of its own and propose a technique to detect the use of frame deletion anti-forensics. The ability of a forensic investigator to detect both editing and the use of anti-forensics results in a dynamic interplay between the forger and forensic investigator. We use develop a game theoretic framework to analyze this interplay and identify the set of actions that each party will rationally choose. Additionally, I will show that anti-forensics can be used protect against reverse engineering. To demonstrate this, I propose an anti-forensic module that can be integrated into digital cameras to protect color interpolation methods
Date: 2/20/2013
Speaker: Yu-Han Yang
Title: Joint Waveform Design and Interference Pre-Cancellation in Time-Reversal Systems
In wideband communication systems, the time-reversal technique can boost the signal-to-noise ratio (SNR) at the receiver with simple single-tap detection. In our previous work, optimal waveform design has been shown to significantly improve the system performance of TR systems. However, when the symbol rate is very high, the severe inter-symbol interference (ISI) still limits the performance at high power region. In this work, we study the joint waveform design and interference pre-cancellation for TR systems. With the proposed joint design, the causal ISI can be subtracted by interference pre-cancellation and the anti-causal ISI can be further suppressed due to the abundant available degrees of freedom. The resulting transmitter structure can be considered as a non-linear waveform design which utilizes the information of previous symbols to refine the signal quality while the receiver structure remains simple. Simulation results demonstrate the remarkable performance improvement over the previous work.
Date: 2/13/2013
Speaker: Yang Gao
Title: Providing Incentives to Microtask Crowdsourcing, Cheaply
Date: 2/6/2013
Speaker: Yi Han
Title: Optimal Waveform Design for Sum Distortion in Multiuser Downlink System
The multi-path effect makes high speed broadband communication a very challenging task due to the severe inter-symbol interference (ISI). By concentrating energy in both the spatial and time domain, time-reversal (TR) transmission technique provide a great potential of low-complexity and energy-efficient communications. However, when the data rate is high, the traditional TR waveforms result in severe ISI. The achievable performance is further degraded in the multiuser downlink system due to inter-user interference (IUI). In this work, we study the weighted sum distortion optimization problem by means of waveform design and power allocation. The optimization problem is finally converted to a sequential optimization problem and the numerical algorithm is shown to converge.
Date: 2/1/2013
Speaker: Hang Ma
Title: State Estimation with Dimension Reduce Preprocessing
State estimation of a system relies heavily on the measurements. With the advance of sensing technology, the ability to measure is no longer a bottleneck in some systems. Engineers are dealing with information-rich estimations. Although information never hurts, how to make the most of the it depends on whether we can process the data efficiently. Sometimes the internal constraint of the system makes it necessary to reduce the dimension of the data before further processing. The problem that the raw measurements are first preprocessed in cluster and then used for estimation was studied. It was shown that there was a lower bound on the required dimension such that if it is beyond the bound, there exists some smart way to design the estimator without incurring any performance degradation. Further, we proposed an algorithm to design the estimator in extreme conditions that even the lower bound can not be satisfied. Different from other algorithms that only ensure convergence to a stationary point, the algorithm ensured convergence close to a global optimal point. The proposed algorithm is illustrated to be effective in three applications.

### Fall 2012

Date: 12/11/2012
Speaker: Xiaoyu Chu
Title: 'Order Detection of Resizing and Contrast Enhancement
Manipulated multimedia content typically undergoes multiple operations. Most existing forensic works are able to detect a specific operation via binary hypothesis test. However, when multiple operations are applied, the order of the operations are more important. It can not only provide a clearer view of the process history, but also implies information about when and who made the forgery. Moreover, the existence of one operation may affect the fingerprints of the other one, which is called conditional fingerprints. In our work, we study the forensic detection of order of operations on a multimedia content. Specifically, we propose a order detection scheme for distinguishing the order of resizing and contrast enhancement. Based on the conditional fingerprints of contrast enhancement given resizing operation, we use Composite hypothesis test tree to make final decisions. In addition, a complementary test is added to the contrast enhancement detection in order to improve the detection performance.
Date: 11/27/2012
Speaker: Qi Wang
Title: Stochastic scheduling & dynamic auction: pricing in the cloud with opportunity cost
This presentation will consist of two part: In the first part, we aim at improving the scheduling mechanism in cloud computing. In the cloud computing environment, usually both the user requests and the computation nodes are highly heterogeneous entities, in terms of service value and service quality. Different matchings can result in vast different performance of the system. Most current cluster scheduling schemes are deterministic, heuristic and controlled by datacenter manager. In this work we make two fundamental changes: 1. From deterministic, heuristic scheduling to a stochastic, optimal scheduling method, where Markov Decision Process is used to learn the optimal stochastic scheduling; 2. The scheduling task is shifted from datacenter manager to a third party broker (auctioneer), who has better incentive of improving utilities for both the users and the datacenter perspectives. Then we show how to use Q-learning to address slow model change in MDP. Built upon this, the second part aims as designing an efficient, incentive compatible and individual rational mechanism for cloud computing market. The efficiency, IC and IR properties are both proved by theory and verified by simulation of the auction process.
Date: 11/20/2012
Speaker: Jungho Myung
Title: Secondary Transmit Design based on Non-Cooperative Feedback Control Game under Sum Feedback-Rate Constraint
It is generally agreed that channel state information (CSI) is of great importance for downlink channel. In practice, however, since the CSI is imperfect due to finite-rate channel feedback, the secondary transmitter is hard to control the interferences to the primary users and among the secondary users in cognitive radio network. In this paper, we propose a noncooperative feedback-rate control game for the self-interested secondary users, in which each user determines the feedback rate to maximize its own utility function. The utility function of each user is defined as the downlink data-rate by CSI feedback size minus the price as a linear function of the CSI feedback rate under sum feedback-rate constraint. The existence of the Nash-equilibrium of the proposed game is analyzed. In order to satisfy the primary users interference threshold and to maximize the sum-rates of the downlink channels, user selection and power control algorithms considering the limited primary users channel are also proposed. Simulation results show that the sum-rate of the proposed game is better than the rate of the equally distributed feedback-size to satisfy the primary users interference constraint.
Date: 11/13/2012
Speaker: Yang Gao
Title: Contract-Based Mechanisms for V2G Ancillary Services: Optimality and Learning
With the foreseeable large scale deployment of electric vehicles (EVs) and the development of vehicle-to-grid (V2G) technologies, it is possible to provide ancillary services to the power grid in a cost efficient way, i.e., through the bidirectional power flow of EVs. A key issue in such kind of schemes is how to stimulate a large number of EVs to act coordinately to achieve the service request. This is challenging since EVs are self-interested and generally have different preferences toward charging and discharging based on their own constraints. In this paper, we propose a contract-based mechanism to tackle this challenge. Through the design of an optimal contract, the aggregator can provide incentives for EVs to participate in ancillary services to the power grid, match the aggregated energy rate with the service request and maximize its own profits. We prove that under mild conditions, the optimal contract-based mechanism takes a very simple form, i.e., the aggregator only needs to publish an optimal unit price to EVs, which is determined based on the statistical distribution of EVs' preferences. We then consider a more practical scenario where the aggregator has no prior knowledge regarding the statistical distribution and study how should the aggregator learn the optimal unit price from its interactions with EVs.
Date: 11/6/2012
Speaker: Feng Han
Title: When femtocells meet time reversal-- TRDMA Cognitive Networks
Today’s tutorial talk consists of two parts. In the first part, a brief introduction about the femtocell will be presented. Specifically, we will cover the following aspects: ? Why does the femtocell gain so much attention? ? What is the femtocell? ? What are the advantages of the femtocell? ? What are the technical challenges of the femtocell? In the second part, motivated by the great potential of time reversal in our recent works, we propose to develop a TRDMA based solution for the next-generation femtocells—TRDMA Cognitive Networks. A systematic overview about this ongoing research will be given and discussed during this talk. Based on what we have accomplished so far, I will focus on three specific aspects about this very promising system: the proposed system architecture (UL and DL), Scheduling and Power Control, and UL interference cancellation. If time permitting, I could show a basic system-level Simulink demo to those who are interested after our group meeting.
Date: 10/23/2012
Speaker: Tong Zhou
Title: Cooperative MIMO Transmission for LTE-A Downlink Using Network Formation Game
In the LTE-A systems, neighboring BSs use the same frequency bands,where inter-cell interference is an important issue. In this talk, we will discuss the design of covariance matrix of transmit signal in MIMO interference channel.Transmitters cooperate with each by doing null shaping to receivers of other links. The generation of the cooperative set is formulated as coalition games and network formation games. The converge and property of the algorithms as coalition games and network formation games are discussed. Some system-level simulation results are shown to evaluate the performance of the proposed algorithms.
Date: 10/16/2012
Speaker: Sha Wei
Title: Relay strategy design and power allocation in energy harvesting cooperative systems
In conventional wireless sensor networks, the transmission is supplied by a pre-charged battery. Even with the design of network lifetime maximization approaches, the sensors still do eventually run out of energy. An alternate emerging solution is the use of energy harvesting (EH) sensors. Specifically, we first introduce the basic energy harvester model and the energy causality constraint. Then, we investigate whether the traditional relay strategies (amplify-and-forward (AF) strategy, decode-and-forward (DF) strategy with link adaptive ratio design and DF with maximum likelihood detection) can achieve full diversity or not under different energy consuming methods. Furthermore, three adaptive and two fixed power allocation strategies are introduced to minimize the error performance and achieve full diversity. From the simulations we can conclude that long-term optimization is better than short-term optimization for its better adaptive ability of CSI. And the system with storage limitation has poorer BER performance than unlimited storage system, which can save more power for later use. The probability of no harvested energy, the initial storage, the maximum storage and the optimization step jointly determine the diversity gain of the system.
Date: 10/9/2012
Speaker: Wei Guan
Title: Two-Way Relay Channel with Opportunistic Network Coding
This work studies the throughput of opportunistic network coding for two-way relay channel. Both static and dynamic network coding are investigated, and it is demonstrated that the network coding gain is upper bounded by some fixed number compared to the conventional relaying. We also distinguish the scenarios with/without delay constraints. It is shown that the end-to-end delay constraint and per-hop delay constraint may beat each other depending on the delay requirement.
Date: 9/25/2012
Speaker: Zhung-Han Wu
Title: Median Filter Anti-Forensics
Median filter operation is a common operation in image manipulation which destroys the trace of some image manipulations. Forensic detection of median filtering on the image would cast doubt on the authenticity of the image. On the other hand, removal of median filtering traces would authenticate the forged images. In this talk, the pixel difference noise is proposed and modeled. A noise addition algorithm for anti-forensics in median filtering is proposed. And the resulting images are tested against known median filtering detectors.
Date: 9/18/2012
Speaker: Yu-Han Yang
Title: Mechanism Design for Wireless Access Network Selection Game
As the proliferation of available wireless access networks such as Wifi access points and femto-cells possibly owned by various operators, network selection becomes an important problem. In the previous talk, we studied which access network a user should connect to in consideration of the rationality and the selfish nature, and we formulated the wireless network selection problem as a stochastic game where the system state consists of the information provided by the networks. However, the networks may not truthfully report the state information if profitable. In this talk, we consider the pricing mechanism design problem with incentive compatibility constraints, which enforce the networks to report the truth, while optimizing the utility of users. The formulated problem is shown to be a mixed integer program which in general has no efficient solution. Exploiting the optimality of substructures, we proposed a dynamic programming algorithm that can optimally solve the problem when the number of networks is 2. For a larger number of networks, the proposed algorithm can perform better than a heuristic greedy approach in a polynomial-time complexity.
Date: 9/11/2012
Speaker: Matt Stamm
Title: Forensic Identification of Order of Processing Operations
Many digital multimedia files undergo some form of processing before they are obtained by an end user. Because processing can alter how multimedia content is perceived and affect how it is used, researchers have developed a number of digital forensic algorithms to verify the authenticity of digital multimedia content. Most existing algorithms are designed to detect the use of a single editing operation. In reality, however, most multimedia content is subjected to several processing operations. In this talk, we examine the problem of forensically determining the order in which processing operations were applied to a multimedia file. We introduce the idea of an operation’s conditional fingerprint and discuss under which scenarios the order of operations can be determined.

### Summer 2012

Date: 8/1/2012
Speaker: Xiangui Kang
Title: Robust Median Filtering Forensics Using Autoregressive Model
One important aspect of multimedia forensics is exposing an image processing history. Median filtering is a popular noise removal and image enhancement tool. It is also an effective tool in anti-forensics recently. As an image is usually saved in a compressed format such as the JPEG format, the forensic detection of median filtering with a JPEG compression remains challenging, because typical filter characteristics are suppressed by JPEG quantization and blocking artifacts. In this paper, we introduce a robust median filtering detection scheme that can overcome such a drawback. Median filtering is first applied on a test image and the difference between the initial image and the filtered output image is called the median filtered residual (MFR) which is used as the forensic fingerprint. Thus, the interference from the image edge and texture, which is regarded as a limitation of the existing forensic methods, can be reduced. Because the overlapped window filtering introduces correlation among the MFRs, an autoregressive (AR) model is proposed to have a concise modeling and its coefficients are used by a Support Vector Machine (SVM) for classification. Experimental results show that the proposed median filtering forensic method is very robust to JPEG compression, with a quality factor as low as 30. It distinguishes well between median filtering and other manipulations and performs well on low-resolution images. The proposed method achieves much better performance than the existing state-of-the-art methods, with very small dimension of features.
Date: 7/25/2012
Speaker: Yan Chen
Title: Optimal Pricing in Stochastic Scalable Video Coding Multicasting System
Heterogeneous multimedia content delivery over wireless networks is an important yet challenging issue. One of the challenges is maintaining the quality of service due to the scarce resource in wireless communications and heavy loadings from heterogeneous demands. A promising solution is combining multicasting and scalable video coding (SVC) techniques via cross-layer design which has been shown to be effectively enhancing the quality of multimedia content delivery service in the literature. Nevertheless, most existing works on SVC multicasting system focus on the static situations, where a snapshot of user demands is given and remains the same. Additionally, the economic value of SVC multicasting system, which is also an important topic from the service provider’s perspective, has seldom been explored. In this work, we study a subscription-based SVC multicasting system with stochastic user arrival and heterogeneous user preferences. A stochastic framework based on Multi-dimensional Markov Decision Process (M-MDP) is proposed for theoretically evaluating the efficiency of the proposed system. A game-theoretic analysis is conducted to understand the rational demands from heterogeneous users under different subscription payment schemes. By transforming the original dynamic and complex M-MDP revenue optimization problem into a traditional MDP problem, we show that the optimal pricing strategy which maximized the expected revenue of the service provider can be derived efficiently. Moreover, the overall user’s valuation on the system, e.g., social welfare, is maximized under such an optimal pricing strategy. Finally, the efficiency of the proposed solutions is evaluated through simulations.
Date: 7/18/2012
Speaker: Hang Ma
Title: Distributed State Estimation in Smart Grid with Communication Constraints
Distributed state estimation in smart grid highly relies on the availability of measurements. Transmitting a lot of measurements within a small time interval is costly and sometimes even impossible. This paper explores the problem of distributed state estimation in smart grid with constraint on the number of measurements that is able to be transmitted in one step. It is shown that there exists a lower bound which depends on the structure of the grid such that if the number of permissible measurements is beyond the bound, then the estimator achieves the same performance as its peer without the constraint. Further, if the number of permissible measurements is below the lower bound, a tradeoff between the performance of the estimator and the measurements transmitted is needed to meet the constraint. A method to attain such tradeoff is offered in this paper. The proposed conclusions and methods are illustrated in the simulation on the IEEE 14-bus system.
Date: 7/11/2012
Speaker: Chunxiao Jiang
Title: Dynamic Chinese Restaurant Game
In this work, we extend the previous Chinese Restaurant Game to the dynamic population setting, where we consider the scenario that the customers arrive and leave by Poisson process. We introduced a Bayesian table state learning method for the customers to estimate whether the table is reserved. We modeled the table selection problem as an MDP problem, proposed a Multidimensional MDP model and a modified value iteration algorithm to find the optimal strategy profile. We further discussed the application of the dynamic Chinese Restaurant Game to the cognitive radio network. Simulation results are shown to verify the effectiveness of the proposed algorithm.
Date: 7/3/2012
Speaker: Han Yi
Title: An Optimal Dynamic Pricing and Schedule Approach in V2G
Smart Grid (SG) can greatly improve the efficiency and reliability of traditional grid. As a promising feature of future SG, the Vehicle-to-Grid (V2G) technique exhibits great potential to balance the supply and demand of electrical power as well as integrate renewable energy. Recently, some V2G-based schemes have been proposed to leverage the energy-storage capability of electric vehicles (EVs) to effectively reduce energy loss caused by supply-demand mismatches. However, most of the existing schemes rely on the assumption that the electricity company is profit-neutral, lacking of adequate incentive to electricity companies for wide deployment. In this paper, we investigate a scenario where the charge station is modelled as an entity driven by its own profit. We formulate the interactions between the charge station and multiple EVs as a game, in which two kinds of EVs, cooperative EVs and selfish EVs, are considered. Regarding the intelligence of selfish EVs, a dynamic pricing over multiple time-slots is developed from the charge station's perspective to maximize its own profit. Both theoretical analysis and simulation results show that through our scheme of dynamic pricing and charge scheduling for cooperative EVs, charge station can maximize its profit.
Date: 6/20/2012
Speaker: Qi Wang
Title: The Voluntary Cloud Computing: a Coalition Game Method for Load Balancing for Distributed Tasks
With the growing popularity of cloud computing, designing high-performance and economically reasonable task scheduling protocols become urgent demand. From the specific perspective of utility computing and voluntary computing, game theory is a suitable tool of protocol design. There are two major contributions in this paper: first we propose a game theoretic approach for solving the load balancing problem for distributive tasks; Second we propose a new business model of voluntary cloud computing, in which the selfishness assumption of cloud providers had abundant rationale. The task scheduling is modeled as a virtual network mapping problem, where the goal is to minimize the maximum congestion (both node congestion and link congestion) in the network. We first propose a centralized method of min-max joint node-edge congestion optimization, which is a mixed integer programming, along with a precise linear programming relaxation method to solve it. Then based on the centralized method we propose a coalition formation game approach with the assumption that each node in the cloud has only the incentive of maximizing its own pay-off. This assumption is well suited for the voluntary computing paradigm that we propose, where an infrastructure like Eucalyptus (a popular cloud computing software) is assumed but all elementary nodes are exposed to the global cloud market. The result of the proposed method is compared with some state-of-the-art methods in terms of several performance indices.
Date: 6/14/2012
Speaker: Yang Gao
Title: Optimal Contract Design for Ancillary Services in Vehicle-to-Grid Networks
With the foreseeable large scale deployment of electric vehicles (EVs) and the development of vehicle-to-grid (V2G) technologies, it is possible to provide ancillary services to the power grid in a cost efficient way, i.e., through the bidirectional power flow of EVs. A key issue in such kind of schemes is how to stimulate a large number of EVs to act coordinately to achieve the service request. This is challenging since EVs are self-interested and generally have different preferences toward charging and discharging based on their own constraints. In this paper, we propose a contract-based mechanism to tackle this challenge. Through the design of an optimal contract, the aggregator can provide incentives for EVs to participate in ancillary services to power grid, match the aggregated energy rate with the service request and maximize its own utility. Simulation results are shown to verify the effectiveness of the proposed contract-based mechanism.

### Spring 2012

Date: 5/8/2012
Speaker: Zhung-Han Wu
Title: Median Filter Anti-Forensics
Multimedia forensics is a promising technique in authenticating an unknown source for the trustworthiness of the multimedia content. On the other hand, multimedia anti-forensics is also crucial in such a way that the weakness of a known detector can be investigated. Our work aims at the anti-forensic of median filter. Median filter distorts the distribution of pixel differences in an image and our work focuses on shifting the statistics of the pixel difference back to an estimate of the original image. We first model the distribution of the pixel differences as generalized Gaussian distribution. After extracting the parameters of the generalized Gaussian distribution, a plausible parameter of distribution of the original image can be estimated. Noise is added to the pixel difference to push the statistics of pixel differences back to the estimated one to conceal the trace of median filtering.
Date: 5/1/2012
Speaker: Tong Zhou
Title: Non-Cooperative Beamforming Game in Downlink Multi-cell Networks
In a multi-cell environment, if a base station transmits data to its own users without considering the interference caused to neighboring cells. The inter-cell interference may limit the transmission rates of all links. In this work, we consider the game theory approach to design the downlink beamformers. Given the target SINR requirement at each links and the transmit power constrain, each transmitter determines its optimal downlink beamforming strategy to maximize its own rate. Numerical results show that proposed game performs better than the present rate game with transmit power constrain only. Both the link average rate and the worst case rate can be enhanced with proper SINR threshold.
Date: 4/24/2012
Speaker: Sha Wei
Title: On the design of relay mapping in multiple access relay channel
For multi-user relay system, the network coding principle has been actively implemented due to its advantage in alleviating the spectral efficiency loss. Previously, in the well-known physical-layer network coding and denoise-and-forward strategies, the relay maps different input signal pairs to the same output message, e.g., both (+1,-1) and (-1,+1) are mapped to +1. Thus, the destination confronts detection ambiguity since it can only depend on the direct link transmission. Consequently, the aforementioned strategies cannot achieve full diversity.To solve such detection ambiguity problem, we propose a new relay mapping strategy for the two-user multiple-access relay channel, where the received signal at the relay is mapped to the 4-ary pulse amplitude modulation signal to avoid the ambiguity issue. Based on the instantaneous relay constellation and destination constellation, we derive the decision regions for different source symbols and obtain the detection error rates at both the relay and destination in closed form, which are shown to be in tight match with Monte-Carlo simulation results. Furthermore, to achieve the full diversity, we implement the link adaptive regenerative (LAR) strategy, where relayed symbols are scaled before being forwarded to the destination. Finally, to minimize the error performances, we further propose a scheme based on maximizing the minimum Euclidean distance of any adjacent constellation points at the destination. Simulations show that our method can achieve similar performances as exhaustive search while greatly simplifying the computations.
Date: 4/17/2012
Speaker: Xiaoyu Chu
Title: Forensic Identification of Compressively Sensed Signals
Due to the ease of tampering digital signals, the authenticity of the acquired signal becomes even more concerned, and lots of forensic techniques have been proposed to adapt to various forgery concerns. Among them, acquisition forensics plays an important role to identify the acquisition process of a signal and verify its authenticity. Recently, a new acquisition technique - compressive sensing - has been broadly studied in many applications for its much lower sampling rate than traditional uniform sampling schemes. Yet, few work has been done in acquisition forensics to take this promising technique into account, i.e., to identify the image been compressive sensed or not. In this talk, we will propose general compressive sensing detectors regardless of what particular signal this technique applies to. Specifically, we propose two detection schemes according to how much information we have about the distribution model of the signal or noise. Moreover, we also propose an estimator for the number of measurements used in compressive sensing. Substantial simulations show our detector and estimator performance with respect to every aspect.
Date: 4/10/2012
Speaker: Wei Guan
Title: On Analysis of Wireless Network-coded Uplink using Non-Coherent Modulations
Network coding has been widely used in the wireless uplink to improve the throughput and provide the spatial diversity. However, the receiver has to estimate the channel coefficients of all users to perform coherent detection, thus the signaling overhead is sometimes formidable and may even outweigh the performance gain. To reduce the channel estimation overhead, we study the non-coherent frequency-shift keying modulations in this work with emphasis on receiver design and performance analysis. Depending on the available channel state information (CSI), we first develop the coherent and non-coherent receivers based on the maximum likelihood (ML) principle for both the analog network coding (ANC) and digital network coding (DNC). As the ML receiver of non-coherent ANC has a non-tractable integral form, we further propose two suboptimum receivers depending on the relative link quality of the source-relay channel and relay-destination channel. We also study the pair-wise error probability of ANC and DNC, and show that the full diversity is still achievable using non-coherent modulations; however, some additional logarithmic term loss would be introduced in the error rate expressions. We demonstrate that the performances loss is much more significant for ANC due to the incapability to efficiently suppress the multi-user interferences. Extensive simulations are also given to verify our analytical results.
Date: 3/6/2012
Speaker: Yu-Han Yang
Title: Admission Control Game in Cellular Networks
In an admission control game, users decide which one of the two cellular cells to enter by maximizing their expected rewards. The immediate reward of each user in a cell decreases as the number of users in the cell increases. When a user aims to maximize his expected reward, his optimal decision depends on the others' decisions. In this work, we formulate the problem of finding the optimal decision rule as a Markov Decision Process (MDP) and propose a modified value iteration algorithm which can efficiently obtain the optimal decision rule. The analysis of the proposed algorithm shows that the optimal decision rule has a simple threshold structure. Numerical simulation results validate the analysis and demonstrate the effectiveness of the optimal decision rule.
Date: 2/28/2012
Speaker: Biling Zhang
Title: Chinese restaurant game in cognitive radio network
In the opportunistic spectrum access model of cognitive radio networks , secondary users (SUs) are allowed to exploit the licensed channels only when the primary users (PUs) are absent. Moreover, SUs usually need to share the available licensed channels. However, due to hardware limitation and imperfect detection, how can the SU obtain accurate information about the true system state, and furthermore, make right decision of accessing a channel to avoid competitions from huge crowds in the same channel, remains a challenge to fully utilize the scare spectrum resources. In this paper, considering a scenario where SUs sense the channels simultaneously and make access decisions sequentially, we model the SUs' decision making process as a Chinese restaurant game. In the proposed game, the SU builds its knowledge of the system state by sensing one of the channels and learning information such as signals and decisions revealed by the previous SUs. It also predicts the decisions of the subsequent SUs to maximize the utility. We analyze the interactions of SUs in the proposed game, specifically, the impact of initial belief, sensing accuracy and the channel quality on SUs' decisions. Some important theoretical conclusions for the two-user two-channel scenario are derived. We finally verify our theoretical conclusions and prove the effectiveness of the proposed scheme through simulation results.
Date: 2/21/2012
Speaker: Feng Han
Title: An Energy-Efficient Switching Strategy for Cellular Network Operations
As we discussed last time, the world-wide cellular networks of today consume a considerable amount of electric power. The energy consumption of cellular networks accounts for near 50% of the operational expenses of cellular service providers. On the other hand, there exists substantial potential through cooperative communications and smarter network operations to save energy. Based on our previous work, we go a step further in our recent work by trying to answer two important questions: 1. What is the condition under which we could switch-off some cells? 2. Which switch-off pattern could maximize overall energy-saving performance under various traffic load demands? In this work, we propose an energy-efficient switching strategy that considers the daily traffic demand profile. We use cooperative communications to extend service from working ("Sun") cells to the switched-off ("Moon") cells with guaranteed quality of service (QoS). We take into account both the call-blocking probability and the combined channel outage probability as the metrics of QoS. Power-saving is maximized among all proposed patterns without compromising the bottom-line QoSs. Both analytical results and numerical evaluations show that the proposed strategy presents great potential of energy saving.
Date: 2/14/2012
Speaker: Yang Gao
Title: Resource Sharing Game with Social Learning
In today’s talk, I will present some recent progress on our resource sharing game project. The resource sharing game is proposed to model the interactions among a group of rational players who share limited resources with unknown statistics, and can find applications in many areas, such as broadband wireless communications, online advertisement and etc. In this game, players will need to predict other players?decisions as well as to learn the statistics of resources in order to maximize their expected utilities. We will show, in this talk, a recursive algorithm that can find an action profile that is a subgame perfect equilibrium of the proposed game. The design of learning rules will also be discussed in today’s presentation.
Date: 2/7/2012
Speaker: Chunxiao Jiang
Title: Renewal-Theoretical Dynamic Spectrum Access in Cognitive Radio Network with Unknown Primary Behavior
Dynamic spectrum access in cognitive radio networks can greatly improve the spectrum utilization efficiency. Nevertheless, interference may be introduced to the Primary User (PU) when the Secondary Users (SUs) dynamically utilize the PU’s licensed channels. If the SUs can be synchronous with the PU’s time slots, the interference is mainly due to their imperfect spectrum sensing of the primary channel. However, if the SUs have no knowledge about the PU’s exact communication mechanism, additional interference may occur. In this paper, we propose a dynamic spectrum access protocol for the SUs confronting with unknown primary behavior and study the interference caused by their dynamic access. Through analyzing the SUs?dynamic behavior in the primary channel which is modeled as an ON-OFF process, we prove that the SUs?communication behavior is a renewal process. Based on the Renewal Theory, we quantify the interference caused by the SUs and derive the corresponding close-form expressions. With the interference analysis, we study how to optimize the SUs?performance under the constraints of the PU’s communication quality of service (QoS) and the secondary network’s stability. Finally, simulation results are shown to verify the effectiveness of our analysis.
Date: 1/31/2012
Speaker: Chih-Yu Wang
Title: Chinese Restaurant Game ?Theory & Applications
In a social network, agents are intelligent and have the capability to make decisions to maximize their utilities. They can either make wise decisions by taking advantages of other agents' experiences through learning, or make decisions earlier to avoid competitions from huge crowds. Both these two effects, social learning and negative network externality, play important roles in the decision process of an agent. We propose a new game, called Chinese Restaurant Game, to formulate the social learning problem with negative network externality. Through analyzing the proposed Chinese restaurant game, we derive the optimal strategy of each agent and provide a recursive method to achieve the optimal strategy. How social learning and negative network externality influence each other under various settings is also studied through simulations. Finally, we illustrate how the proposed framework can be applied to many problems in different research fields, such as wireless networking, cloud computing and online social networking.

### Fall 2011

Date: 11/29/2011
Speaker: Xiaoyu Chu
Title: Forensics of Compressive Sensing on (Nearly) Sparse Signal Acquisition
Compressive sensing is a promising technique which realizes a sub-Nyquist sampling rate for sparse and nearly sparse signals. Many applications have been studied and implemented for all kinds of sparse signals, such as images, radar, cognitive radio and so on, either for lower sampling or denoising purpose. However, little work has been done on acquisition forensics regarding this new technique. In this talk, I'll present our work on compressive sensing forensics to distinguish those noisy sparse or nearly sparse signals obtained by compressive sensing from those acquired using uniformly Nyquist sampling technique. Specifically, we categorize the feasible signals for compressive sensing into noisy sparse and nearly sparse signals. And propose the detection scheme for each model under different assumptions of the signal knowledge we got. In addition, the estimation of number of measurements is also proposed for noisy sparse signal with simulation done to see the performance.
Date: 11/22/2011
Speaker: Wei Guan
Title: On The Study of Multi-Source Amplify-And-Forward Network
We study the error performance of the multiple-access relay channel using amplify-and-forward protocol. Both of the variable gain relaying (VGR) and fixed gain relaying (FGR) are studied, where the relay nodes are subject to short-term power constraint and long-term power constraint, respectively. For the single-relay system, we show that both of VGR and FGR can achieve a diversity order of (2,-1), and the pairwise error probability (PEP) increases linearly with the number of sources (i.e., K). When there are multiple relays (i.e, L), we propose a relay selection scheme to minimize the maximum PEP, and show that the diversity order is (L+1,-L). To improve the coding gain, we then study the distributed space-time block coding (DSTBC). We prove that DSTBC-FGR can always achieve full primary diversity, whereas the diversity of DSTBC-VGR is upper bounded by min (L+1,K+1). For the single-source network, we then develop a selective DSTBC-VGR scheme by adaptively allocating the relay power, and we show that the diversity is (L+1,0). Finally, we study the diagonal distributed space-time coding (DDSTC). We show that the diversity is (L+1,-L) for both of DDSTC-VGR and DDSTC-FGR, and the code design criterion is to maximize the minimum product distance.
Date: 11/8/2011
Speaker: Yan Chen
Title: Clustering using Coalition Formation Game Modeling and Submodular Optimization
Date: 11/1/2011
Speaker: Matthew Christopher Stamm
Title: Information Security Through Anti-Forensics
Over the past decade, researchers have developed a variety of digital forensic techniques to verify the authenticity and trace the processing history of digital multimedia files. While forensic techniques have been traditionally viewed as a tool to prevent and detect the falsification of digital multimedia content, this does not mean that they cannot be used for malicious purposes. For example, digital forensic techniques can possibly be used to reverse engineer proprietary signal processing components inside digital devices. In this talk, we show how anti-forensics can be used to provide information security in situations when forensic tools can be abused. Specifically, we propose an anti-forensic module that can be integrated into a digital camera’s signal processing pipeline to prevent the reverse engineering of its color filter array and color layer interpolation technique.
Date: 10/25/2011
Speaker: Yu-Han Yang
In this work, we consider the non-codebook-based precoder design problem for multi-user MIMO downlink with per antenna power constraints in LTE-Advanced systems. The Maximum SINR filter is applied to maximize the criterion of sum SINR for the jointly encoded multiple data streams within a user. It has been shown that such a criterion makes the joint decoding become close to ML decoding. The proposed algorithm utilizes Lagrangian duality to solve for the precoders with the primal-dual method. Simulation results demonstrate that the proposed algorithm significantly outperforms other conventional methods such as zero-forcing (ZF), block-diagnalization (BD), and signal-to-leakage-and-noise ratio (SLNR).
Date: 10/4/2011
Speaker: Xiangui Kang
Title: Scalable lossy compression for encrypted image
Compression of encrypted data is becoming an active research topic in recent years. The traditional way of securely and efficiently transmitting redundant data is to first compress the data to reduce the redundancy, and then to encrypt the compressed data to mask its meaning. At the receiver side, the decryption is performed before prior to decompression. However, in some application scenarios, a sender needs to transmit some data to a receiver and hopes to keep the information confidential to a network operator who provides the channel resource for the transmission. The sender should encrypt the original data and the network provider always has interest in reducing the data rate, and may tend to compress the encrypted data without any knowledge of the cryptographic key and the original data. At receiver side, a decoder integrating decompression and decryption functions will be used to reconstruct the original data. In this talk, a novel scheme for scalable lossy compression for encrypted image using stream cipher and the primary results will be presented.
Date: 9/27/2011
Speaker: Wangrok Oh
Title: A New Channel Reliability Value for Iterative MAP Turbo Decoder
Several previous works have demonstrated that to optimize the performance of iterative maximum a posteriori probability (MAP) decoder for turbo codes, it is crucial to estimate the channel signal-to-noise ratio (SNR) within the iterative MAP turbo decoder. In this talk, we investigate the behavior of the iterative MAP turbo decoder under the channel SNR estimation error and propose a new channel reliability value (CRV) for iterative MAP turbo decoder. Unlike the traditional CRV, the proposed CRV is not a function of channel SNR. Hence, it is not required to estimate the channel SNR within the iterative MAP turbo decoder and thus it simplifies the decoder implementation. Simulation results show that the iterative MAP decoder employing the proposed CRV provides better error performance compared to the traditional iterative MAP decoder.
Date: 9/20/2011
Speaker: Feng Han
Title: Energy-Efficient Cellular Network OPeration via Base Station Cooperation
The rising cost of energy and increased environmental awareness have sparked a keen interest in the development and deployment of energy-efficient communication technologies. The energy efficiency of cellular networks can be increased significantly by selectively switching off some of the base stations (BSs) during periods of low traffic load. In this paper, we propose a scalable BS switching strategy and use cooperative communication techniques and power control to extend network coverage to the service areas of the switched-off BSs. The outage probability and the achievable power savings of the proposed scheme are analyzed, both analytically and numerically, and a potential of up to 50% power saving is observed in the numerical results.

### Summer 2011

Date: 8/24/2011
Speaker: Biling Zhang
Title: Cooperation Stimulation for Dynamic Spectrum Access Networks
Dynamic spectrum access is a promising technique enabling the efficient usage of the limited spectrum. In the scenario when a primary user is far away from the base station and is experiencing low achievable rates while some secondary users half way between the primary user and the base station and has more favorable channel conditions, how can we utilize the spectrum so that we can improve the QoS of PU while assign SU the chance to access the vacant spectrum? In this work, we propose a novel spectrum sharing model considering the fair and efficient allocation of both used and unused channels and incorporate indirect reciprocity game to stimulate the cooperation between primary and secondary users to improve the overall system performance. We develop a reputation updating policy under time-vary channel and show the existence of stationary reputation distribution for any given optimal action rule. We also theoretically derive the stable condition of the optimal action rule.
Date: 8/17/2011
Speaker: Yang Gao
Title: Cooperation Stimulation for Multiuser Cooperative Communication Networks Using Indirect Reciprocity Game
The viability of cooperative communications largely depends on the willingness of users to help. However, in future wireless networks where users are rational and pursue different objectives, they will not help to relay information for others unless this can improve their own utilities. Therefore, it is very important to study the incentive issues when designing cooperative communication systems. In this paper, we propose a cooperation stimulation scheme for multiuser cooptative communication networks using indirect reciprocity game. By introducing the notion of reputation and social norm, users care about their future utilities and get the incentive to cooperate with others. Different from existing works on reputation based schemes that mainly relay on experimental verifications, we theoretically demonstrate the effectiveness of our scheme in two steps. First, we conduct steady state analysis of the game and show that cooperating with users having good reputation can be sustained as an equilibrium. Then, by modeling the action spreading at the transient state as an evolutionary game, we show that the equilibriums we found are stable and can be reached with proper initial conditions. Moreover, we deal with possible cheating behaviors of users through the introduction of energy detection. Finally, simulation results are shown to verify the efficiency and effectiveness of the proposed scheme.
Date: 8/10/2011
Speaker: Chih-Yu Wang
Title: Chinese Restaurant Game in Social Networks: Initial Findings
Social learning, which agents learn knowledge from other agents' behaviors and shared information, is a trending topic in the research of social networks. In traditional social learning problem, the payoff of a player depends only on his own action and the unknown state. The externality from other players' actions is seldom discussed, especially the negative effect due to competition from other players. In this work, we proposed the Chinese restaurant game as a new game to analyze users¡¯ competition behaviors in social networks with social learning characteristics. In this early report, we analyzed the equilibrium in simultaneous game with perfect/imperfect signals. We also derived the best response and equilibrium in sequential game in recursive form.
Date: 8/3/2011
Speaker: Matt Stamm
Title: Cheat-Proof Demand Response Games for Smart Grids
To prevent the spread of digital forgeries, a variety of digital forensic techniques have been developed to detect manipulations made to multimedia content. In response, both forgers and forensics researchers have developed a number of anti-forensic operations designed to fool digital forensic techniques. This situation is further complicated by the fact that anti-forensic operations can leave detectable traces in digital content, allowing forensic investigators to develop techniques capable of detecting the use of anti-forensics. As a result, digital forgers and forensic investigators must respond to each other¡¯s actions in order to achieve their goals (i.e. creating undetectable forgeries and detecting digital forgeries respectively). In this talk, we investigate the interplay between the actions of a forger and forensic investigator. We discuss how to evaluate the performance of forensic and anti-forensic algorithms in the presence of opposing forensic and anti-forensic techniques. Additionally, we use game theory to identify the optimal strategies employed by the forger when using anti-forensics and the forensic investigator when attempting to detect forgeries. We use our recently proposed video frame deletion forensic and anti-forensic algorithms to perform a case study of these new techniques.
Date: 7/27/2011
Speaker: Wei Guan
Title: Mitigating Error Propagation in Wireless Network-Coded Uplink
Network coding has been widely used in wireless uplink to improve spectral efficiency. However, the relay decoding error may propagate to the intended receiver and thus severely degrade the diversity performance. So in this work, we develop two power scaling schemes at the relay side and two detection schemes at the receiver side, respectively, to mitigate error propagation and thus achieve full diversity. For the soft power scaling based link adaptive relaying, we develop a virtual source-relay-destination channel model and demonstrate that the relay power should be such to balance the signal-to-noise ratios of the source-relay channel and relay-destination channel. As for the hard power scaling based ON-OFF relaying, we first design a decision rule based on total pairwise error probability, and then simplifies it to the threshold-based relaying strategy. At the receiver side, we show that the weighted minimum distance detection with the weight being determined by the relative link quality of source-relay channel and relay-destination channel can achieve full diversity once the global channel state information is available, otherwise the maximum likelihood detection that explicitly takes into account relay decoding error should be employed to achieve full diversity. Finally we compare these schemes in terms of relay power consumption and signalling overhead, and validate our results through simulations.
Date: 7/20/2011
Speaker: Xiaoyu Chu
Title: Image forensics in compressive sensing
Image forensic algorithms of tracing an image's processing history has been well studied regarding traditional compression, like JPEG and JPEG2000. However, little consideration has been given to a recent technique - compressive sensing, with which we can acquire the compressed data by directly sensing the objects, like single-pixel camera. In this talk, we show that compressive sensing, like other traditional compression methods, also leave fingerprint on the data of the image in wavelet domain. We also propose a two step detection to distinguish the fingerprint come from different compressions, and further tell whether an image has been compressive sensed.
Date: 7/13/2011
Speaker: Wan-Yi Lin
Title: Game-Theoretical Behavior Analysis for Social Networks
Within the past decade, the proliferation of multimedia social network communities, such as Napster, and YouTube where millions of users form a dynamically changing infrastructure to share content, have introduced the new concept of social networking that creates a technological revolution as well as brings new experiences to users. The massive content production poses new challenges to the scalable and reliable sharing of multimedia content over large and heterogeneous networks. It also raises critical issues of intellectual property protection and privacy issues. In a multimedia social network, users actively interact with each other, and such user dynamics not only influence each individual user but also affect the system performance. To provide a predictable and satisfactory level of service, it is of ample importance to understand the impact of human factors on multimedia social networks. Such an understanding provides fundamental guidelines to the better design of multimedia systems and networking, and offers more secure and personalized services. For example, in a peer-to-peer file-sharing system, users pool together the resources and cooperate with each other to provide an inexpensive, highly scalable, and robust platform for distributed data sharing. However, since the nature of participation in many multimedia social networks is often voluntary and unregulated, there is a need to provide incentives and mechanism to stimulate cooperation among users to improve system performance. This talk will illustrate various aspects of issues and problems in multimedia social networks via two case studies of human behavior in multimedia fingerprinting, peer-to-peer live streaming, and mobile video sharing.
Date: 6/22/2011
Speaker: Liang Xiao
Title: Jamming-Resistant Collaborative Broadcast In Wireless Networks
Date: 6/15/2011
Speaker: Yan Chen
Title: Cheat-Proof Demand Response Games for Smart Grids
While demand response has achieved promising results on making the power grid more efficient and reliable, the additional dynamics and flexibility brought by demand response also increase the uncertainty and complexity of the centralized load forecast. In this talk, we propose a game theoretic demand response scheme that can transform the traditional centralized load prediction structure into a distributed load prediction system by the participation of customers. Moreover, since customers are generally rational and thus naturally selfish, they will cheat if cheating can improve their payoff. Therefore, enforcing truth-telling is crucial. We prove analytically and demonstrate with simulations that the proposed game theoretic scheme is cheat-proof, i.e., all customers are motivated to report and consume their true optimal demands and any deviation will lead to a utility loss.
Date: 6/10/2011
Speaker: Dr. Anthony Kuh
Title: Renewable Energy and Island Sustainability (REIS) Program and Stochastic Models of Micro-grids
This talk discusses starts by discussing Hawaii's energy landscape and the REIS program. The state of Hawai'i and the US Department of Energy (DOE) signed a memorandum of understanding in 2008 known as the Hawaii Clean Energy Initiative (HCEI) requiring 70% of Hawaii's energy come from clean sources by 2030. At the beginning of 2009 we formed a multidisciplinary research and education program at the University of Hawai'i at Manoa, (REIS) to focus on renewable energy, sustainability, and smart grids. The REIS group won an internal UHM sustainability competition in 2009 and in 2010 won a DOE work force training grant in the Strategic Training and Education in Power Systems. We discuss goals of our group including building the REIS program, research and education activities, and interfacing with other UHM sustainability efforts, outside academic institutions, industry, and government. We then discuss some recent work in the smart grid area. We propose modeling micro-grid parameters using probabilistic formulations. A goal is to model the micro-grid beyond the last substation including distributed generation sources (wind and solar). The micro-grid will have measuring devices such as AMI at some locations. We model the micro-grid as a factor graph and use belief propagation to perform inference on the grid. A goal is to estimate parameters using message passing and then use this information to perform control and security on the micro-grid.

### Spring 2011

Date: 5/2/2011
Speaker: Zhenzhen Gao
Title: Multi-source cooperative transmission with network coding
A new cooperative transmission protocol and network coding scheme for synchronized multi-source system are proposed. The users are organized in two groups. Users in separate groups transmit in separate time slots. Users in the same group transmit with orthogonal waveforms. In each transmission, the user transmits his own information and relays other users¡¦ information at the same time. The proposed protocol is able to mimic the full-duplex transmission with half-duplex mode to achieve high data rate. The network code for this protocol can be designed using classical linear block code. It is shown that the minimum Hamming distance of the network code determines the system diversity order. Therefore, with well-designed network code, the proposed protocol can achieve high diversity order. Simulation results demonstrate its superior performance compared with previous work.
Date: 4/25/2011
Speaker: Yang Gao
Title: Enabling Cooperative Peer-to-Peer Streaming Using Mechanism Design
Cooperative peer-to-peer (P2P) streaming, which enables cooperation among geographically neighboring peers with large intra-group bandwidth, is regarded as a promising approach to address the network inefficiency problem encountered by the traditional non-cooperative P2P schemes. However, due to the selfish nature of peers, they may tend to report false private information in order to act as free-riders, which consequently leads to a substantial system inefficiency. In this talk, we propose a mechanism design based cooperative P2P streaming scheme which achieves the optimal system-wide streaming performance. In particular, with properly designed transfer function, we show that it is a dominant strategy for each user to participate in the cooperative streaming and report their private information truthfully. Moreover, in the proposed scheme, the transfer is proved to exchange only within the group without surplus or deficit at any time.
Date: 4/18/2011
Speaker: Feng Han
Title: An Energy Efficient Wideband Green Communication Paradigm—introducing a Multi-band Time Reversal Structure

Wireless Green Communication technology has long been desired with today’s fast expansion of wireless infrastructures and applications throughout the world. In seeking of novel wireless green communication technologies, we discovered that the time reversal (TR) transmission technique can be a promising candidate, due to its inherent capability of efficiently harvesting signal energy from the multipath environment and its low complexity.

In this talk, we take a closer look at the essence of the time reversal structure and its fundamental theoretical limits. As will be shown in the first part of this talk, the time reversal structure tends to be dedicated to system reliability by maximizing the energy-efficiency and diversity order (through time-domain processing), compared with its counterpart OFDM (mostly frequency-domain processing) in favor of high multiplexing order as another extreme.

Based on this observation, it would be interesting and important to investigate whether a desirable tradeoff between the two benefits can be obtained by an innovative system design. In the second part, we introduce a multi-band time reversal structure that can achievable arbitrary tradeoff between the benefits that are otherwise exclusively possessed by single-band TR and OFDM. Besides, the multi-band structure allows added flexibility, such as multi-hopping, multiple access, resource allocation, security and regulatory spectrum operation.

To summarize, the time-reversal (single/multiple band structures) is an ideal candidate for (very) wideband wireless green communications where energy-efficiency, reliability and interfering/informational leakage are the primary concerns, compared with the spectrum efficiency.

Date: 4/11/2011
Speaker: Steve Tjoa
Title: Sparse and Nonnegative Factorizations for Music Understanding

Constrained matrix factorization has become widely popular for performing tasks in music information retrieval (MIR) such as music transcription and source separation. However, the basic formulations of these factorization algorithms have trouble decomposing particular types of musical data. Recent developments in factorization techniques that exploit the sparsity or nonnegativity of data have improved the effectiveness of these factorizations. Nevertheless, the sophistication and efficiency of these methods must keep up with the growing size and complexity of the data around us.

In this dissertation, I propose methods for sparse and nonnegative factorization that is specifically suited for analyzing musical signals. First, I discuss two additional constraints that aid factorization of musical signals: harmonic and co-occurrence constraints. Harmonic dictionary learning imposes additional harmonic constraints upon the atoms of the learned dictionary while allowing the dictionary size to grow appropriately during the learning procedure. Our proposed method restricts each spectral atom to contain only one musical pitch. When there is significant spectral-temporal overlap among the musical sources, our method outperforms popular existing matrix factorization methods as measured by the recall and precision of learned dictionary atoms. Meanwhile, co-occurrence constraints enforce dependence within predetermined groups of atoms, thereby allowing objects to be represented using multiple atoms instead of only one atom. We introduce three simple and convenient multiplicative update rules for NMF that enforce dependence among atoms. Using examples in music transcription, we demonstrate the ability of these updates to represent each musical note with multiple atoms and cluster the atoms for source separation purposes.

Second, we study how spectral and temporal information extracted by nonnegative factorizations can improve upon musical instrument recognition. Musical instrument recognition has drawn plenty of attention for its ability to facilitate other tasks in music information retrieval such as transcription and search. However, instrument recognition in melodic musical signals remains difficult, especially for classification systems that rely entirely upon spectral information instead of temporal information. Here, we propose a simple and effective method of combining spectral and temporal information for instrument recognition. While existing classification methods use traditional features such as statistical moments, we extract novel features from spectral and temporal atoms generated by nonnegative matrix factorization (NMF) using a biologically motivated multiresolution gamma filterbank. Unlike other methods that require thresholds, safeguards, and hierarchies, the proposed spectral-temporal method requires only simple filtering and a flat classifier. This method is also effective; instrument recognition accuracy in monophonic isolated sounds is 92.3% for twenty-four instrument classes, and accuracy in solo melodic phrases is 96.2% for eleven instrument classes. Using instrument family classifications, accuracy in solo melodic phrases is 97.4% for three instrument families.

Finally, we study how to perform sparse factorization when a large dictionary of musical atoms is already known. Sparse coding methods such as matching pursuit (MP) have been applied to problems in music information retrieval such as transcription and source separation with moderate success. However, when the set of dictionary atoms is large, identification of the best match in the dictionary with the residual is slow -- linear in the size of the dictionary. Here, we propose a variant called approximate matching pursuit (AMP) that is faster than MP while maintaining scalability and accuracy. Unlike MP, AMP uses an approximate nearest-neighbor (ANN) algorithm to find the closest match in a dictionary in sublinear time. One such ANN algorithm, locality-sensitive hashing (LSH), is a probabilistic hash algorithm that places similar, yet not identical, observations into the same bin. We illustrate the effectiveness of AMP by decomposing musical signals with respect to a large dictionary containing over 170,000 spectra of musical sounds. While the accuracy of AMP is comparable to similar MP methods, the computation time is reduced. Also, by using LSH, this method scales easily; the dictionary can be expanded without reorganizing any data structures.

Date: 4/4/2011
Speaker: Matt Stamm
Title:
Date: 3/28/2011
Speaker: Quoc Lai
Title: Wireless Network Cocast: Cooperative communications with space-time network coding
Traditional cooperative communications can greatly improve communication performance. However, transmissions from multiple relay nodes are challenging in practice. Single transmissions using time-division multiple access cause large transmission delay, but simultaneous transmissions from two or more nodes using frequency-division multiple access (FDMA), code-division multiple access (CDMA), or distributed space-time codes are associated with the issues of imperfect frequency and timing synchronization due to the asynchronous nature of cooperation. In this dissertation, we propose a novel concept of wireless network cocast (WNC) and develop its associated space-time network codes (STNCs) to overcome the foretold issues. In WNC networks, each node is allocated a time slot for its transmission and thus the issues of imperfect synchronization are eliminated. To reduce the large transmission delay, each relay node forms a unique signal, a combination of the overheard information, and transmits it to the intended destination. The combining functions at relay nodes forms a STNC that ensures full spatial diversity for the transmitted information as in traditional cooperative communications. Various traditional combining techniques are utilized to design the STNCs, including FDMA-like and CDMA-like techniques and transform-based techniques with the use of Hadamard and Vandermonde matrices. However, a major distinction is that the combination of information from different sources happens within a relay node instead of through the air as in traditional cooperative communications. We consider a general case of multiuser relay wireless networks, where user nodes transmit and receive their information to and from a common base node with the assistance from relay nodes. We then apply the STNCs to multiuser cooperative networks, in which the user nodes are also relay nodes helping each other in their transmission. Since the cooperative nodes are distributed around the network, the node locations can be an important aspect of designing a STNC. Therefore, we propose a location-aware WNC scheme to reduce the aggregate transmit power and achieve even power distribution among the user nodes in the network. WNC networks and its associated STNCs provide spatial diversity to dramatically reduce the required transmit power. However, due to the additional processing power in receiving and retransmitting each other's information, not all nodes and WNC networks result in energy efficiency. Therefore, we first examine the power consumption in WNC networks. We then offer a TDMA-based merge process based on coalitional formation games to orderly and efficiently form cooperative groups in WNC networks. The proposed merge process substantially reduces the network power consumption and improves the network lifetime.
Date: 3/14/2011
Speaker: Nathan Goergen
Title: EXTRINSIC CHANNEL-LIKE FINGERPRINT EMBEDDING FOR TRANSMITTER AUTHENTICATION IN WIRELESS SYSTEMS
To secure next-generation wireless systems, robust authentication mechanisms are required to mitigate acts of message forgery and the nefarious actions of malicious actors. While cryptographically secure authentication devices have historically been included in upper-layer wireless protocols, these approaches require that nodes impart a significant amount of upper-layer processing to extract and decode the authentication message. Authentication at the physical-layer can prevent wasteful processing of unintended, uninteresting, or maliciously fabricated transmissions, allowing nodes to quickly authenticate legitimate users and implicate charlatans. We propose a fingerprinting method for authenticating wireless systems that exploits typical receiver preprocessing algorithms, such as channel equalization, to ameliorate the distortions of previous "blind superposition" embedding approaches. We consider augmenting current intrinsic channel-based authentication schemes with an! extrinsic, synthetically generated fingerprint signal that is embedded by the transmitter. We consider the heterogeneous wireless broadcast system, consisting of modified (aware) and unmodified (unaware) receivers, while also considering scenarios where channel state information (CSI) is available / unavailable to the transmitter. We present a taxonomy of fingerprinting approaches that operate in the time, frequency, and spacial dimensions, and we discuss how our method can be applied to multiple-input-multiple-output (MIMO) and orthogonal frequency division multiplexing (OFDM) systems. Finally, we demonstrate how our fingerprinting method can be used in Dynamic Spectrum Access theaters, by establishing edifice for active spectrum sensing. While the focus of this work is on signal fingerprinting for the purpose of transmitting an authentication signal, our embedding approach is not limited to authentication applications.
Date: 2/28/2011
Speaker: Yan Chen
Title: Multimedia Social Networks: Game Theoretic Modeling and Equilibrium Analysis
Multimedia content sharing and distribution over multimedia social networks is more popular now than ever before: we download music from Napster, share our images on Flickr, view user-created video on YouTube, and watch peer-to-peer television using Coolstreaming, PPLive and PPStream. Within these multimedia social networks, users share, exchange, and compete for scarce resources such as multimedia data and bandwidth, and thus influence each other's decision and performance. Therefore, to provide fundamental guidelines for the better system design, it is important to analyze the users' behaviors and interactions in a multimedia social network, i.e., how users interact with and respond to each other. Game theory is a mathematical tool that analyzes the strategic interactions among multiple decision makers. It is ideal and essential for studying, analyzing, and modeling the users' behaviors and interactions. In this thesis, game theory will be used to model users' behaviors in social networks and analyze the corresponding equilibria. Since users in different multimedia social networks may have different types of interdependency, to effectively model the users' behaviors and interactions, we should use different games for different multimedia social networks. In this thesis, we first illustrate how to use game theory to analyze and model users' behaviors in multimedia social networks by discussing the following three scenarios: 1. In the first scenario, we consider a non-cooperative multimedia social network where users in the social network compete for the same resource. We use multiuser rate allocation social network as an example for this scenario. 2. In the second scenario, we consider a cooperative multimedia social network where users in the social network cooperate with each other to obtain the content. We will use cooperative peer-to-peer streaming social network as an example for this scenario. 3. Since users in social networks are rational thus naturally selfish, they will not cooperate with others unless cooperation can improve their own performance. Therefore, cooperation stimulation is an important issue in cooperative social networks. In the third scenario, we consider how to use the indirect reciprocity game to stimulate cooperation among users. Moreover, the concept of multimedia social networks" can be applied into the field of signal and image processing. If each pixel/sample is treated as a user, then the whole image/signal can be regarded as a multimedia social network. From such a perspective, we introduce a new paradigm for signal and image processing, and develop generalized and unified frameworks for classical signal and image problems. In this thesis, we use image denoising and image interpolation as examples to illustrate how to use game theory to re-formulate the classical signal and image processing problems.
Date: 2/21/2011
Speaker: Yu-Han Yang
Utilizing channel reciprocity, the traditional time-reversal technique boosts the signal-to-noise ratio at the receiver with very low transmitter complexity. However, the large delay spread gives rise to severe intersymbol interference (ISI) when the data rate is high, and the performance is further degraded in the multiuser downlink due to the inter-user interference (IUI). With perfect channel state information (CSI), optimum linear waveform design can be calculated to satisfy each user's quality-of-service (QoS) constraint. In practice, however, estimation errors are inevitable and thus the estimated channel is imperfect. Such imperfect CSI can impair the performance severely due to the sub-optimal design. Therefore, in this work, we explored the robust waveform design for the time-reversal multiuser downlink communication system. We formulated an optimization problem with MSE QoS constraints and considered the true channel to be in a set of channels with bounded error from the estimated channel. The optimization problem is further transformed into a standard convex optimization which can be efficiently solved using common optimization packages. Numerical simulation results are shown to demonstrate the proposed robust waveform design can outperform the non-robust one at high SNR.
Date: 2/14/2011
Speaker: Wei Guan
Title: Clustering Based Many-To-One Wireless Network Cocast
In this talk, we propose a new protocol that leverages user cooperation to improve the transmission reliability to a common destination. The whole data frame is divided into K time slots, in which all clusters work in a TDMA way. Besides sending local symbols, each node also behaves as a decode-and-forward relay to other clusters. The local symbol and relay symbol are linearly coded by the signature waveforms to avoid inter-user interferences. Transmit beamforming is used within each cluster to guarantee coherent combining at the destination, and the decoding at both the source nodes and destination is based on linear de-correlator. We derived both the exact Symbol Error Rate (SER) for M-ary PSK signal and the asymptotic SER at high SNR regions. We showed that the transmission delay can be reduced dramatically with a linear loss of diversity order. Finally, the simulation results justify the advantage of the proposed protocol.
Date: 2/7/2011
Speaker: Wan-Yi Lin
Title: Game-Theoretical Formulation for Image Prediction
In this talk I will present a game-theoretical approach to provide a framework for optimal template selection in image prediction. Image prediction is an effective tool for coding still images and intra pictures in video. Template matching algorithms which use neighboring blocks of the prediction target as templates have been widely used for image prediction. The assumption of these approaches is that the template has a similar textural structure as the prediction target. Up to now these approaches all use pre-fixed templates for all prediction target. However, in real images, these fixed templates are very likely to contain textures that are not or are not significant in the prediction target and these insignificant textures introduces larger prediction residues or cause failure of template matching. In this paper, we propose a coalitional game in which every pixel is treated as a player and tries to seek partners to form a coalition to capture the texture structure. By forming a coalition, every player in the coalition can obtain a gain of improving the ability of capturing the texture structure of coalition while incurring a cost of introducing textural variance within the coalition. Experimental results show that the proposed game-theoretical approach outperforms the conventional pre-fixed template matching prediction up to 2dB coding gain.
Date: 1/31/2011
Speaker: Quoc Lai
Title: Coalition Formation Games for Energy Efficient Wireless Network Cocast
Cooperative communications, in which single-antennas nodes cooperate to each others such as those in wireless network cocast (WNC), can achieve transmit power saving. However, due to additional processing power in relaying each other information, not all nodes and cooperative network results in energy efficiency. To ensure energy efficiency in WNC networks, we propose a merge process that is based on coalition formation games to form cooperative groups. A node is merged into a cooperative group if the merge leads to power saving for the group and also the individual members. Simulation is provided to corroborate the energy efficient WNC.

### Fall 2010

Date: 12/1/2010
Speaker: Nate Goergen
Title: Improving Channel-Like Fingerprinting Overlays using Predicted Channel State Information
A physical-layer fingerprint embedding scheme for orthogonal frequency division multiplexing (OFDM) transmissions using an overlay approach is presented, where the fingerprint signal conveys a low capacity communication suitable for authenticating the transmission and further facilitating secure communications. We improve upon channel-like fingerprinting overlay designs that blindly use past channel state information by leveraging predictive filtering to reduce model mismatch error when recovering the fingerprint signal. The detection performance of a digital fingerprint message in time-varying channel conditions is presented, where CSI used to design the overlay is obtained via Kalman prediction. A fingerprint signal bit error rate improvement of 8-9 dB over designs that do not incorporate prediction is demonstrated through simulation for some channel conditions.
Date: 11/24/2010
Speaker: Zhenzhen Gao
Title: Differential Cooperative Communicatons with Space-Time Network Coding
In multinode cooperative communications, simultaneous transmissions from two or more nodes are challenging due to its asynchronous nature. In addition, channel estimation is a costly task due to the amount of training, especially when the number of cooperative users is large. Considering these practical challenges in multinode cooperative communications, we propose a new transmission scheme, namely differential space-time network coding (DSTNC), to overcome the problems of imperfect synchronization and complicated channel estimation. Each user in the network linearly combines the correctly decoded symbols by a network coding vector, which is designed to achieve full diversity without introducing large time delay. The pairwise error probability (PEP) is analyzed and the design criteria of the DSTNC are derived based on the PEP. The proposed DSTNC scheme can be applied to any number of cooperative users. Simulation results are shown to verify the performance of the proposed transmission scheme.
Date: 11/10/2010
Speaker: Yan Chen
Title: Evolutionary Games for Image Interpolation
In this talk, we study the image interpolation from the game theoretic perspective and formulate the image interpolation problem as an evolutionary game. In this evolutionary game, the players are the unknown high resolution pixels and the pure strategies of the players are the corresponding low resolution neighbors. By regarding the non-negative weights of the low resolution pixels as the probabilities of selecting the pure strategies, the problem of estimating the high resolution pixels becomes finding the evolutionarily stable strategies for the evolutionary game. Experimental results show that the proposed game theoretical approach can achieve better performance than the state-of-the-art image interpolation methods in terms of both PSNR and visual quality.
Date: 11/3/2010
Speaker: Matt Stamm
Title: Anti-Forensics for Frame Deletion/Addition in MPEG Video
Due to the ease with which digital information can be altered, many digital forensic techniques have recently been developed to authenticate multimedia content. One important digital forensic result is that adding or deleting frames from an MPEG video sequence introduces a temporally distributed fingerprint into the video can be used to identify frame deletion or addition. By contrast, very little research exists into anti-forensic operations designed to make digital forgeries undetectable by forensic techniques. In this talk, we propose an anti-forensic technique capable of removing the temporal fingerprint from MPEG videos that have undergone frame addition or deletion. We present experimental results that demonstrate that our proposed anti-forensic technique can effectively remove this fingerprint.
Date: 10/27/2010
Speaker: Yongle Wu
Title: Thwarting Selfish and Malicious Behaviors in Cognitive Radio Networks: A Game Theoretic Approach

We first develop a cheat-proof repeated spectrum sharing game, which provides the incentive for selfish users to cooperate with each other and reveal their private information truthfully. We propose specific cooperation rules based on the maximum total throughput and proportional fairness criteria, and investigate the impact of spectrum sensing duration on system performance.

We also consider the situation where a group of selfish users collude for higher payoffs. We propose a novel multi-winner spectrum auction framework which did not exist in auction literature, and develop collusion-resistant auction mechanisms to suppress the collusive behaviors. In addition, we apply the semi-definite programming relaxation to significantly reduce the complexity of algorithms.

When malicious users are taken into consideration, we apply game theoretic tools to thwart potential malicious behavior in cognitive radio networks. Specifically, we model the anti-jamming defense as a zero-sum game, and derive the optimal strategy for secondary users to execute in face of jamming threats. Moreover, we propose learning schemes for secondary users to gain knowledge of adversaries.

Finally, we consider security countermeasures against eavesdroppers, and propose a cooperation paradigm that primary users improve secrecy with the help of trustworthy secondary users. We derive the achievable pair of primary users' secrecy rate and secondary users' transmission rate under various circumstances, and model the interaction between primary users and secondary users as a Stackelberg game.

Date: 10/20/2010
Speaker: Yang Gao
Title: Image Interpolations with Adaptive Covariance Estimation
Image interpolation addresses the problem of generating a high-resolution image from its low-resolution version. Edge-directed interpolation schemes, which can adapt to local structures through the assistance of high-resolution image covariance, achieve better visual quality than traditional interpolation schemes. However, these methods usually assume stationarity in pixel values and estimate covariance based on some heuristically determined window, which is not optimal. Motivated by this fact, in this talk, we will discuss the possibility of adaptive estimation of image covariance and its application in image interpolation. Moreover, we demonstrate, by experiment results, the potential performance of this method and illustrate some key issues of applying coalition formation game to this problem.
Date: 10/13/2010
Speaker: Steve Tjoa
Title: Cross-Layer Musical Search
Content-based search is one of the most central problems in music information retrieval. This type of search uses an audio query, often provided as an acoustic waveform, to retrieve similar pieces of music from a database. Operations such as audio fingerprinting and cover song retrieval have been used to search depending upon the level of similarity desired between the query and the retrieved matches. One type of search -- an acoustic waveform query into a symbolic MIDI database -- has been overlooked. Considering the abundance of MIDI data openly available, this type of search is growing in importance. However, the mapping of acoustic and symbolic data into a common feature space has not been thoroughly investigated for the purpose of search. In this talk, we propose a system that executes an efficient search using approximate nearest-neighbor methods as well as a common feature set that is robust to musical and environmental variations between the acoustic and symbolic domains.
Date: 10/6/2010
Speaker: Feng Han
Title: Time-Reversal Division Multiplexing in Multi-path Channels
Multi-path effect makes high speed broadband communications a very challenging task due to the severe inter-symbol interference (ISI). By concentrating energy in both the spatial and temporal domains, time-reversal (TR) pre-filtering technique provides a great potential of low-complexity energy-efficient communications. In this paper, we propose to use TR pre-filtering in a multiuser downlink system with large delay spread channels, and develop a new concept of time-reversal division multiplexing (TRDM), in which the location-specific signatures between the base station and users are used to separate intended signals. We investigate the system performance in the cases with one and multiple transmit antennas, and the latter can provide an even better spatial focusing effect. Experimental measurements are shown to verify the spatial and temporal focusing effects in real-life multi-path environments. We also use numerical simulations to justify the theoretical analysis, and the proposed scheme is shown to achieve good performance.
Date: 9/15/2010
Guest speaker
Date: 9/8/2010
Speaker: Yu-Han Yang
Title: Sum Rate Optimization for Time-Reversal Multiuser Downlink Systems with Linear Precoding
Utilizing channel reciprocity, the time-reversal (TR) technique boosts the signal-to-noise ratio at the receiver with very low transmitter complexity. However, the large delay spread gives rise to severe intersymbol interference (ISI) when the data rate is high, and the performance is further degraded in the multiuser downlink due to the inter-user interference (IUI). In this work, we study the weighted sum rate optimization problem in the multiuser downlink with ISI using linear precoding. By exploiting the relation between the allocated power and the SINR targets, an iterative algorithm is proposed to jointly optimize the power allocation together with the transmit precoder design. Simulation results are shown to demonstrate the superior performance of the proposed optimization algorithm in comparison with TR and the zero-forcing method.

### Summer 2010

Date: 8/18/2010
Speaker: Nate Goergen
Title: Extrinsic Channel-Like Fingerprinting Overlays Using Channel Side-Information
We present a method for overlaying orthogonal frequency division multiplexing (OFDM) transmissions with a fingerprint message at the physical-layer, where the fingerprint signal conveys a low capacity communication suitable for authenticating the transmission and further facilitating secure communications. A novel approach is discussed where previous channel-state information (CSI) is incorporated into the design of the fingerprint, allowing the signal to mimic distortions typical of time-varying channels. The fingerprint message is only visible to aware receivers who explicitly preform detection of the signal, but is invisible to receivers employing typical channel equalization. A taxonomy of overlay designs is discussed and these designs are explored through experiment using time-varying CSI recorded from an IEEE802.16e Mobile WiMax base-station. The performance of the fingerprint signal as received by a WiMax subscriber is demonstrated using CSI measurements derived from the downlink signal. Detection performance for the digital fingerprint message in time-varying channel conditions is also presented via simulation.
Date: 8/11/2010
Speaker: Wei Guan
Title: Two Way Relaying with Non-Coherent DBPSK Modulation: Receiver Design and Performance Analysis
The traditional half-duplex relaying system with coherent modulation is suffering low channel use and increased channel estimation overhead. To effectively overcome these shortcomings, this work focuses on a two-way relaying protocol with non-coherent differential modulation. The bidirectional communications, assisted by a set of K parallel relays, take place in two phases. In the multiple access (MAC) phase, the two sources simultaneously send their differentially encoded BPSK symbols to the relays; while in the broadcast (BC) phase, all relays send through orthogonal channels a differentially encoded symbol that indicates whether the two source symbols have the same signs or not. We designed the optimal maximum a posteriori (MAP) detectors at both the relays and sources. To analyze the error performance, we approximated the MAP detector at the relays as a multi-user detector followed by a network coding encoder. In the case of single relay, we showed that the MAP decoder at the sources is actually equivalent to the standard DBPSK decoder for the relay-to-source channel and thus derived the exact end-to-end decoding error. To minimize the average decoding errors, we also investigated the power allocation problem by use of asymptotic analysis. We showed that the optimal source power is inversely proportional to the square root of the channel gain of the source-to-relay channel, and the optimal relay power decreases with system SNR. For the multiple relay cases, though the exact analysis is unavailable, we developed the upper and lower bounds on system performances and showed that the diversity order is the ceiling of k/2.
Date: 8/4/2010
Speaker: Zhenzhen Gao
Title: Weighted Space-Time Network Coding for Secure Cooperative Communications with Asynchronous Transmissions
Due to the broadcast nature, wireless transmissions can be overheard by any receiver near the destination. A physical layer approach for secure wireless cooperative communications is proposed in this paper. Considering the asynchronous nature of cooperative communications, this paper proposes a weighted space-time network coding (WSTNC) scheme for multinode cooperative communications to prevent eavesdropping and overcome the problem of imperfect synchronization. In the proposed scheme, training symbols are transmitted by the destination (D) instead of the user nodes. Therefore, each user node can obtain the channel state information (CSI) between itself and D, while the eavesdropper (E) can get nothing useful. By exploiting local CSI and the symbols to be transmitted, weighting coefficient is designed for each user node to prevent E from detecting and ensure successful detection at D. Space-time network codes are design to achieve full diversity at D and improve the transmission efficiency for the asynchronous network. The design criteria of the WSTNC is derived based on the security requirement and the pairwise error probability (PEP). Simulation results are presented to verify the performance of the proposed WSTNC scheme.
Date: 7/28/2010
Speaker: Feng Han
Title: TR-based Multicasting Networks: Possibility and Performances
Time-reversal (TR) is an efficient pre-equalization technique with proven capability of leveraging the channel signature property and collecting multipath gain in large delay spread wireless environment. A large amount of literatures and experiments investigated its performances and temporal-spatial focusing effect in the single user scenario. Yet, little work has been done in exploiting the special focusing effect in the multi-user networks. Motivated by that, in this work, we proposed a TR-based multicasting communication system with highly affordable complexity and worked in hoping of being able to answer the following fundamental questions: Is it possible (or feasible) to do so? What is the SINR achieved at each user on average? How fast can this network communicate? What is the outage performance of this network? In this talk, we would first looked at a system model that reveals the essence of multi-path wireless channels but is still analytically tractable for us to get meaningful results. And then, we shall together evaluate its SiNR level, maximum achievable sum rate, and three different kinds of outage probabilities. Analytical expressions and equations will be presented, but detailed derivation will be skipped in this talk due to limited time. A set of numerical results will also be shown at the end of this talk. A stronger generalization equipped with multiple antennas is currently ongoing and almost done, which will come out shortly but not be included in today’s talk.
Date: 7/21/2010
Speaker: Yang Gao
Title: Image Interpolation Games
Image interpolation addresses the problem of generating a high-resolution image from its low-resolution version. Most existing methods of image interpolation mainly rely on local information, which fail when the local geometry cannot be inferred from the neighboring pixels. Motivated by recent development of non-local methods and game theoretic modeling in image denoising, we propose a new scheme for image interpolation. The basic idea is to first interpolate the missing pixels by a classical algorithm, e.g. bilinear, which can be regarded as noisy observations of true values. Then, based these noisy observations and known pixels, we apply linear adaptive filter to non-locally recover the true values of missing pixels. In this recovery process, the design of the optimal structure of candidate sets is critical, which we solve by coalition formation game. Finally, preliminarily experimental results will be shown and followed by the discussion of the future work.
Date: 7/14/2010
Speaker: Beibei Wang
Title: Asynchronous Cooperative Sensing for Dynamic Spectrum Access
Cognitive radio technology has been receiving a growing attention in recent years as a revolutionary communication paradigm that can utilize the existing wireless spectrum resources more efficiently. In order to fully utilize spectrum resources, it is very important to design efficient dynamic spectrum access schemes, and one critical prerequisite is precise detection of primary user’s appearance to avoid interference. Various spectrum sensing schemes have been proposed, such as energy detector, feature detector, and matched filter detector. To utilize spatial diversity, multi-user cooperative spectrum sensing is also proposed, which has been shown to achieve better detection accuracy. However, most of the existing cooperative spectrum sensing schemes assume that the secondary users will measure and report the primary user’s presence at the same time, which may actually result in collisions with the primary user. In today’s talk, an asynchronous cooperative spectrum sensing is presented, in which different users sense the licensed spectrum band and fuse their local detection results to the fusion center at different time epochs. The fusion center can adopt an optimal decision combination structure so that the average cost of sensing can be minimized. It is shown that the asynchronous cooperative sensing achieves lower cost than synchronous cooperative sensing.
Date: 7/7/2010
Speaker: Wan-Yi Lin
Title: Physical layer fingerprinting for spectrum sensing
Cognitive Radio is a promising technology that can alleviate the spectrum shortage problem by enabling unlicensed users equipped with cognitive radios to coexist with incumbent users in licensed spectrum bands while causing no interference to incumbent communications. Spectrum sensing is one of the essential mechanisms of cognitive radios and its operational aspects are being investigated actively. Till now the effort has been made by the secondary users for spectrum sensing. However, the primary user may have the incentives to help the secondary users to improve the security of CRs as well as reduce the interference of primary transmission. In this talk, we will demonstrate the tradeoff of primary user between the degradation of legacy receiver and reduce the false positive of spectrum sensing by applying physical layer fingerprint onto the primary signal.
Date: 6/30/2010
Speaker: Yongle Wu
Title: A Secrecy Game in Cognitive Radio Networks
Although cognitive radio technology improves efficiency of spectrum utilization, usually primary users do not gain directly from opening up the spectrum except spectrum trading. However, if information secrecy is a concern, primary users may benefit from the simultaneous transmission of secondary users. In this talk, we investigate the situation where an eavesdropper attempts to decode a primary user's message, and the primary user improves the secrecy rate with the help of a secondary user. In particular, we derive the pair of the primary user's secrecy rate and the secondary user's transmission rate under various circumstances, and further model this scenario as a game where both users decide how much power to use. The Nash equilibrium of this game is analyzed, and simulation results are presented to verify the performance.
Date: 6/23/2010
Speaker: Yan Chen
Title: Some Details of Image Denoising Games
How to adaptively choose optimal neighborhoods is very important to pixel-domain image denoising algorithms since too many neighborhoods may cause over-smooth artifacts and too few neighborhoods may not be able to efficiently remove the noise. While the Stein’s principle is shown to be able to estimate the true mean square error (MSE) for determining the optimal neighborhoods, there exists a trade-off between the accuracy of the estimate and the minimum of the true MSE. In this paper, we study the impact of this trade-off and formulate the image denoising problem as a coalition formation game. In the game, every pixel is treated as a player, who tries to seek partners to form a coalition to achieve better denoising results. By forming a coalition, every player in the coalition can obtain a gain of improving the accuracy of the Stein’s estimate while incurring a cost of increasing the minimum of the true MSE. Moreover, we propose a heuristically distributed approach for coalition formation. We also show that the traditional approaches that use a heuristically determined candidate set are special cases of the game theoretical framework by choosing the utility function without a cost term. Finally, experimental results show that the proposed game theoretical approach can achieve better performance than the nonlocal method in terms of both PSNR and visual quality.
Date: 6/16/2010
Speaker: Steve Tjoa
Title: How to Find Side Information for Music Understanding
Music information retrieval is hard -- often too hard to perform without making some assumptions. If we design solutions that cover the entire space of acoustic possibilities, which researchers often do, we may design a system that is either inaccurate or prohibitively slow. One way to impose meaningful assumptions is through side information: relevant information that eases the task of processing a particular input. Simply put, side information makes many unsolvable tasks in music information retrieval solvable. But how do we find this side information? One valuable source is the MIDI file. Thankfully, there are millions of MIDI files all over the Internet waiting to be harvested. However, finding the correct side information for a given input is not straightforward. In this talk, we propose the idea of a "cross-layer search" -- in this case, finding a MIDI file that relates to a given musical waveform. By performing a content-based search on a large MIDI database, we can efficiently find side information for MP3 or WAV files and therefore improve the performance of subsequent MIR systems.
Date: 6/9/2010
Speaker: Matt Stamm

### Spring 2010

Date: 5/6/2010
Speaker: Yang Gao
Title: Image Quality Assessment Games
Objective image quality metric plays a critical role in many image processing applications. The most widely used quantity, mean square error (MSE), however, has been shown to be not able to fully predict human perception of image fidelity and quality. Recently, the structural similarity (SSIM) index which captures the structural change between images has been proposed and proved to be effective. Theoretical comparisons between MSE and SSIM motivate us to calculate SSIM using structural similar pixels instead of involving all pixels within a pre-defined block heuristically. We then formulate the image quality assessment problem as a coalition formation game. In this game, each pixel is treated as a player who seeks partners to form a coalition to achieve better assessment. By forming a coalition, every player in the coalition can obtain a gain of reducing the variance of the estimate while incurring a cost of increasing the bias. Finally, preliminarily experimental results will be shown and followed by the discussion of the future work.
Date: 4/29/2010
Speaker: Yu-Han Yang
Title: Improving on Time-Reversal with Interference Suppression
Time reversal precoding maximizes the signal-to-noise ratio (SNR) by matched filtering the multipath channel. However, it results in severe temporal sidelobes which cause the inter-symbol-interference (ISI) and thus error floor at high SNR. The intimate connection between multipath and multiple antennas enable us to formulate the multipath channel filter design as a multiple-input multiple-output (MIMO) beamformer design problem. Several different criteria are considered. We show that Minimum mean square error (MMSE) and maximum signal-to-interference-and-noise ratio (MSINR) precoding filters mitigate the ISI while minimizing the degradation of SNR. Simulation results are presented to illustrate the performance improvement.
Date: 4/22/2010
Speaker: Feng Han
Title: Time-Reversal Multicasting
This talk presents what ideas we have in mind about TimeReversal-based multicasting. It is necessay to point out that this is not yet a complete work. Instead, it investigates the possibilities of applying Time-reversal to multicast application and things beyond. This presentation will be highly conceptual, but tackle some critical issues in TR-based multicasting networks that allow the system to go from theory to reality. In this talk, some basic observations for multicasting will be discussed, which motivate the proposed TR-based multicast system. Also, a effective family of waveform is proposed to mitigate inter-user interference and enhance physical-layer security. Some claims are made togeher with supporting simulation results.
Date: 4/15/2010
Speaker: Zhenzhen Gao
Title: Inter-symbol Interference Analysis for Time Reversal Transmission
The future communication systems will have to accommodate larger data rates, the natural solution consisting of increasing the bandwidth brings a number of open challenges. In this matter, time reversal (TR) could play a role in the design of the broadband wireless communications. In TR, the time reversed version of the channel impulse response (CIR) is used to precode the transmitted symbols. These time reversed waves propagate in the channel, retrace their former paths and eventually lead to a focus of power in space and time at the receiver. Despite that the channel energy is focused in time and space, the TR system cannot erase inter-symbolinterference (ISI) completely. In this work we examine the effect of ISI by comparing the average energy of ISI and symbol. We derive the exact expressions for the first- and second-order moments of ISI and symbol for TR transmission using the simplified SV channel model. The upper-bound of average ISI to signal ratio is given for SV model. The analytical results can be used to investigate the effects of channel parameters and signal interval on the average ISI to signal ratio.
Date: 4/8/2010
Speaker: Nate Goergen
Title: Extrinsic Channel-Like Fingerprinting for Authenticating MIMO Systems
A framework for introducing an extrinsic fingerprint signal to space-time coded transmissions at the physical-layer is presented, whereby the fingerprint signal conveys a low capacity digital communication suitable for authenticating the transmission and further facilitating secure communications. A novel approach is discussed where the fingerprint signal mimics distortions similar to time-varying channel effects. Specifically, the fingerprint signal is only visible to aware receivers considering previous channel state information and is otherwise invisible to a receiver equalizing according to current channel state information. An augmented signal is created consisting of the original transmission, or primary message, and the fingerprint message. Two example fingerprint signals and detection rules are presented based on pulse-amplitude keying and phase-shift keying approaches. The methods for obtaining the real (intrinsic) channel estimate, the extrinsic fingerprint message, and the primary transmission are analytically demonstrated using general pilot embedding schemes. The worst-case distortions caused by non-ideal equalization of a fingerprinted message are derived using the 2x2 Alamouti code. Simulation results including bit error rate (BER) and model mismatch error using a maximum-likelihood (ML) receiver are presented for both the primary and fingerprint signal, while authentication signal BERs lower than the primary signal are demonstrated.
Date: 4/1/2010
Speaker: Wan-Yi Lin
Title: Optimal pricing for mobile video streaming networks
Mobile phones are among the most popular consumer devices and the recent developments of 3G networks and smart phones enable users to watch video programs by subscribing data plans from service providers. Due to the ubiquity of mobile phones and phone-to-phone communication technologies, subscribers can redistribute the video content to non-subscribers. Such a redistribution mechanism is a potential competitor for the service provider and is very difficult to trace given the users' high mobility. The service provider has to set a reasonable price for the data plan to prevent such re-distribution behavior to protect his/her own profit. In this paper, we analyze the optimal price setting for the service provider by investigating the equilibrium between the subscribers and the secondary buyers in the content-redistribution network. We model the behavior between the subscribers and the secondary buyers as a seller-buyer game and find the optimal price and quantity for both groups of users. We use the optimal price of the redistribution networks to determine the price of the content that can either prevent the redistribution or optimize the content owner's utility. Such an analysis can help the service provider preserve their profit under the threat of the redistribution networks and can improve the quality of service for end users.
Date: 3/25/2010
Speaker: Yan Chen
Title: Image Denoising Games
How to adaptively choose optimal neighborhoods is very important to pixel-domain image denoising algorithms since too many neighborhoods may cause over-smooth artifacts and too few neighborhoods may not be able to efficiently remove the noise. While the Stein's principle is shown to be able to estimate the true mean square error (MSE) for determining the optimal neighborhoods, there exists a trade-off between the accuracy of the estimate and the minimum of the true MSE. In this paper, we study the impact of this trade-off and formulate the image denoising problem as a coalition formation game. In the game, every pixel is treated as a player, who tries to seek partners to form a coalition to achieve better denoising results. By forming a coalition, every player in the coalition can obtain a gain of improving the accuracy of the Stein's estimate while incurring a cost of increasing the minimum of the true MSE. Finally, experimental results show that the proposed game theoretical approach can achieve better performance than the nonlocal method in terms of both PSNR and visual quality.
Date: 3/11/2010
Speaker: Matt Stamm
Title: Anti-Forensics of JPEG Compression
The widespread availability of photo editing software has made it easy to create visually convincing digital image forgeries. To address this problem, there has been much recent work in the field of digital image forensics. There has been little work, however, in the field of anti-forensics, which seeks to develop a set of techniques designed to fool current forensic methodologies. In this work, we present a technique for disguising an image’s JPEG compression history. An image’s JPEG compression history can be used to provide evidence of image manipulation, supply information about the camera used to generate an image, and identify forged regions within an image. We show how the proper addition of noise to an image’s discrete cosine transform coefficients can sufficiently remove quantization artifacts which act as indicators of JPEG compression while introducing an acceptable level of distortion. Simulation results are provided to verify the efficacy of this anti-forensic technique.

Title: Forensic Estimation and Reconstruction of a Contrast Enhancement Mapping
Due to the ease with which convincing digital image forgeries can be created, a need has arisen for digital forensic techniques capable of detecting image manipulation. Once image alterations have been identified, the next logical forensic task is to recover as much information as possible about the unaltered version of image and the operation used to modify it. Previous work has dealt with the forensic detection of contrast enhancement in digital images. In this paper we propose an iterative algorithm to jointly estimate any arbitrary contrast enhancement mapping used to modify an image as well as the pixel value histogram of the image before contrast enhancement. To do this, we use a probabilistic model of an image’s pixel value histogram to determine which histogram entries are most likely to correspond to contrast enhancement artifacts. Experimental results are presented to demonstrate the effectiveness of our proposed method.
Date: 3/4/2010
Speaker: Yongle Wu
Title: Time-Reversal Architecture: Theory and Practice
Traditional communication techniques may not perform well under a severe multi-path channel. A novel method is the so-called time reversal, which essentially implements a matched filter without complicated structures. The transmitter uses the time-reversed channel response as a basic waveform, and the energy will be automatically focused at the receiver. In this talk, we will give a general overview of the time-reversal architecture, both the theoretic background and practical issues related to the project. Details of software simulation and hardware implementation are discussed, and simulation results are presented to gain an insight of this novel approach.
Date: 2/25/2010
Speaker: Beibei Wang
Title: Anti-Jamming Stochastic Game for Cognitive Radio Networks
In this paper, we investigate the security mechanism when secondary users are facing the jamming attack, and propose a stochastic game framework for anti-jamming defense. At each stage of the game, secondary users observe the spectrum availability, the channel quality, and the attackers?strategy from the status of jammed channels. According to this observation, they will decide how many channels they should reserve for transmitting control and data messages and how to switch between the different channels. Using the minimax-Q learning, secondary users can gradually learn the optimal policy, which maximizes the expected sum of discounted payoffs defined as the spectrum-efficient throughput. The proposed stationary policy in the anti-jamming game is shown to achieve much better performance than the policy obtained from myopic learning, which only maximizes each stage’s payoff, and a random defense strategy, since it successfully accommodates the environment dynamics and the strategic behavior of the cognitive attackers.
Date: 2/18/2010
Speaker: Quoc Lai
Title: Space-Time Network Codes with Transform-based Coding
Traditional cooperative communications can improve communication reliability. However, transmissions from multiple relay nodes are challenging in practice. Single transmissions in TDMA manner cause large transmission delay, but simultaneous transmissions from two or more nodes are associated with the issue of imperfect frequency and timing synchronization. In this talk, following the novel concept of wireless network cocast (WNC) presented previously, transform-based coding is utilized to design space-time network codes for a group a N client nodes that cooperate to one another in their transmissions to a common base node. Each client node also acts as a relay to form a unique linearly-coded signal from the overheard symbols of other source nodes and transmit it to the base node in its dedicated time slot. Based on the pair-wise error probability (PEP), the code matrix is designed to achieve full diversity for all transmitted symbols.
Date: 2/4/2010
Speaker: Steve Tjoa
Title: Musical Instrument Recognition from Temporal Nonnegative Matrix Factorization Atoms
Instrument recognition has drawn plenty of attention for its ability to facilitate other tasks in music information retrieval such as transcription and search. This classification problem has been adequately solved for monophonic musical signals but not for polyphonic signals. Furthermore, most classification systems rely entirely upon spectral information instead of temporal information. In this work, we test the hypothesis that temporal information can improve upon the accuracy achievable by the state of the art in polyphonic musical instrument recognition. Unlike existing temporal classification methods which use traditional features such as temporal moments, we extract novel features from temporal atoms generated by nonnegative matrix factorization (NMF) in a biologically motivated manner. We show that this proposed feature, the multiresolution gamma filterbank response (MGFR), can differentiate among certain families of instruments, and when combined with spectral information, can improve classification performance even further.

### Fall 2009

Date: 12/9/2009
Speaker: Steve Tjoa
Title: Musical Instrument Recognition using Temporal Envelope Parameters and Nonnegative Matrix Factorization
Musical instrument recognition has recently benefited from advancements in dictionary learning. Because musical signals contain acoustic events that often reoccur over many time units, the data in a musical signal is highly redundant. There already exist dictionary-learning methods that exploit the spectral redundancy among sounds in a musical signal. Many of these methods depend upon nonnegative matrix factorization (NMF) to obtain low-rank approximations of audio spectrograms of the signal. While these methods are effective in removing spectral redundancy from the signal, redundancy remains in the temporal domain. The temporal activities of factorized atoms usually resemble known patterns, particularly in musical signals. In this presentation, we propose a method for classifying musical signals that exploits these common patterns of temporal activity. From each atom, we extract a set of temporal envelope parameters which quantify the attack rate, decay rate, and energy modulation of the temporal envelope. When combined with spectral features, these parameters improve classification performance over methods which only rely on spectral information.
Date: 12/2/2009
Speaker: Yan Chen
Title: Indirect Reciprocity Game Modelling For Cooperation Stimulation In Cognitive Networks
In cognitive networks, since nodes generally belong to different authorities and pursue different goals, they will not cooperate with others unless cooperation can improve their own performance. Thus, how to stimulate cooperation among nodes in cognitive networks is very important. However, most of existing game-theoretic cooperation stimulation approaches rely on the assumption that the interactions between any pair of players are long-lasting. When this assumption is not true, according to the well-known Prisoner's Dilemma and the backward induction principle, the unique Nash equilibrium (NE) is to always play non-cooperatively. In this paper, we propose a cooperation stimulation scheme for the scenario where the number of interactions between any pair of players are finite. The proposed algorithm is based on indirect reciprocity game modelling where the key concept is "I help you not because you have helped me but because you have helped others". We formulate the problem of finding the optimal action rule as a Markov Decision Process (MDP) and propose a modified value iteration algorithm to find the optimal action rule. Using the packet forwarding game as an example, we show that with an appropriate cost-to-gain ratio, the strategy that forwarding $i$ packets to the receiver with reputation level i is an evolutionarily stable strategy (ESS). Finally, simulations are shown to verify the efficiency and effectiveness of the proposed algorithm.
Date: 11/25/2009
Speaker: Zhenzhen Gao
Title: Differential space-time network coding for cooperative communications
Much research in cooperative communications has focus on simultaneous transmissions from different relay nodes with an assumption of perfect timing and frequency synchronization and perfect channel information at the receivers, which is challenging to be met in practical transmissions. Recent work have investigated the design of distributed space-time code to overcome the issue of imperfect timing or/and frequency synchronization. Exact channel state information is required at all the receivers in these schemes. However, channel estimation is known to be a challenging and costly task especially in multi-antenna or multi-node wireless systems since the amount of training or convergence time (incurred by blind techniques) grows with the number of links. It’s imperative to develop differential techniques for multi-node cooperative communications since their burden of channel estimation is even more severe than multi-antenna systems. In this talk, a differential space-time network coding scheme will be introduced, based on space-time network code(STNC) proposed by Lai et al.. The new coding scheme can overcome the synchronization and channel estimation problems, and it differs from STNC in two aspects: differential encoding and method of network coding.
Date: 11/18/2009
Speaker: Yu-Han Yang
Title: Multi-user MIMO downlink beamforming with SINR constraints
In this work, we aim to solve the multiuser multi-input multi-output (MIMO) downlink beamforming problem where one multi-antenna base station broadcasts data to many users. Each user is assigned multiple data streams and has multiple antennas at its receiver. The data streams for each user are subject to an average signal-to-interference-plus-noise-ratio (SINR) constraint. Efficient solutions to the joint transmit-receive beamforming and power allocation problems based on iterative methods are proposed. The adopted beamforming method is a maximum SINR filter bank which exploits cooperation between the data streams of a user. Based on this filter bank, we derive an SINR balancing property which simplifies the complex power allocation problem to a linear one. Simulation results verify the superiority of the proposed algorithms over previous works with approximately the same complexity.
Date: 11/11/2009
Speaker: Feng Han
Title: On Energy Spreading Transform Based MIMO Systems: Capacity and Diversity-Multiplexing Tradeoff
Very recently, an energy spreading transform (EST) based iterative detection scheme was proposed to multiple-input-and-multiple-output (MIMO) fading channels, ending up with an EST-based MIMO system. The EST spreads the energy of each symbol over the entire symbol block, providing diversity and enhancing the detection reliability at the receiver. Most of existing analyses on this EST-based MIMO system mainly focus on its bit-error-rate (BER) performance. However, more fundamental aspects such as capacity has not been well investigated. In this talk, I shall first show our evaluation on the contribution of the EST-based iterative processing to the system capacity. And then I would present the optimal trade-off between diversity and multiplexing of this EST-based MIMO system. Some counterintuitive results will be shown for further discussion in hoping to solicit some helpful suggestions from you.
Date: 11/4/2009
Speaker: Nate Goergen
Title: Best-Effort Cooperation Using Non-Dedicated Relays
Traditional cooperative communications consider dedicated relays, while often such relays may not be available. We consider wireless transceivers that cooperatively relay signals in addition to their own primary communication missions. Under the best-effort delivery policy, a node is not obligated to devote energy for relaying signals, nor does it provide a guarantee of signal quality to retransmissions. Instead the relay sacrifices energy at its own discretion with priority given to the primary communication mission. We consider one best-effort delivery problem: a system that helps transmit an additional signal within its original transmission energy budget while inducing minimal degradation to the primary signal. To maintain this constraint, we consider the feasibility of reallocating energy from pilot signals used for channel estimation toward the relaying service, as channel conditions become stationary. We demonstrate that transmitter energy may be strategically allocated between a relay component and a pilot component of the transmission using best-effort delivery, and that this allocation may dynamically change in response to changing channel conditions and/or QoS requirements. We present the dynamic power allocation problem using generalized pilot-embedding methods, to demonstrate the possibility of best-effort transmission with minimal degradation to primary transmissions. Optimal power allocation rules with respect to primary-user are discussed.
Date: 10/28/2009
Speaker: Yongle Wu
Title: The Optimal Defense of a Secondary User Against Jamming Attacks
Cognitive radio technologies have become a promising approach to increase the efficiency of spectrum utilization. Although cognitive radio has been intensively studied in recent years, only a few works have discussed security aspects. In this talk, we focus on the jamming attack, one of major threats to cognitive radio networks, where several malicious attackers want to jam the secondary user's communication link by injecting interference. We model this scenario into a jamming game, in which the optimal strategy can be derived using the Markov decision process approach. Furthermore, learning schemes are proposed for the secondary user to learn the wireless environment, such as the primary users' access pattern and the number of attackers. Finally, simulation results are presented to verify the performance.
Date: 10/21/2009
Speaker: Matt Stamm
Title: Forensic Estimation and Reconstruction of a Contrast Enhancement Mapping
Due to the ease with which convincing digital image forgeries can be created, a need has arisen for digital forensic techniques capable of detecting image manipulation. Once image alterations have been identified, the next logical forensic task is to recover as much information as possible about the unaltered version of image and the operation used to modify it. Previous work has dealt with the forensic detection of contrast enhancement in digital images. In this talk we propose an iterative algorithm to jointly estimate any arbitrary contrast enhancement mapping used to modify an image as well as the pixel value histogram of the image before contrast enhancement. To do this, we use a probabilistic model of an image’s pixel value histogram to determine which histogram entries are most likely to correspond to contrast enhancement artifacts. Experimental results are presented to demonstrate the effectiveness of our proposed method.
Date: 10/14/2009
Speaker: Wan-Yi Lin
Title: Cooperation stimulation strategies for Peer-To-Peer wireless live video-sharing social networks
Users watching live streaming in the same wireless network share the same limited bandwidth of backbone connection to the Internet, thus they might want to cooperate with each other to obtain better video quality. These users form a wireless live-streaming social network. Every user wishes to watch video with high quality while paying as little as possible cost to help others. This paper focuses on providing incentives for user cooperation. We propose a game-theoretic framework to model user behavior and to analyze the optimal strategies for user cooperation simulation in wireless live streaming. We first analyze the Pareto optimality and the time-sensitive bargaining equilibrium of the two-person game. We then extend the solution to the multiuser scenario. We also consider potential selfish users?cheating behavior and malicious users?attacking behavior and analyze the performance of the proposed strategies with the existence of cheating users and malicious attackers. Both our analytical and simulation results show that the proposed strategies can effectively stimulate user cooperation, achieve cheat free and attack resistance, and help provide reliable services for wireless live streaming applications.
Date: 9/30/2009
Speaker: Beibei Wang
Title: Anti-Jamming Defense in Cognitive Radio Networks Using Stochastic Game Modeling
Cognitive radio is an emerging communication paradigm in recent years that has received an increasingly growing attention, since it can utilize the limited spectrum resources in a more intelligent and flexible way. Since the radio spectrum environment is dynamic in nature, due to primary users?activity, a lot of stochastic modeling approaches, such as those based on Markov chain and Markov Decision Process (MDP) modeling, have been proposed so as to predict future traffic patterns. As various secondary users compete for the spectrum resources, non-cooperative games are also considered to model the intelligent interactions among secondary users. However, few of these works have considered the existence of malicious users who will launch harmful attacks to a secondary user network, and provide a quantitative analysis of their potential damage. In today’s presentation, I will talk about one prevailing attack in wireless networks, jamming attack, and analyze its impact on a! cognitive radio network. I will first formulate the anti-jamming defense of the secondary users against malicious users as a (zero-sum) game. Considering the dynamic spectrum opportunities, I extend the static game to a stochastic game, where the secondary users and malicious users?actions together with the primary users?activities determine the state transitions. Then, I derive the optimal policy for both players using minimax-Q learning, which only requires each player’s own local information for updating strategies. Finally, the optimal policy for a simple cognitive radio network setting is demonstrated through simulations, followed by conclusions and future work.
Date: 9/23/2009
Speaker: Steve Tjoa
Title: Two Improvements to Nonnegative Matrix Factorization for Music Transcription: Co-occurrence and Harmonic Constraints
Music source separation has drawn plenty of attention for its ability to facilitate tasks in music information retrieval. However, source separation is difficult because there does not exist a unique solution; given a musical mixture, there are several valid ways to decompose the mixture into its individual sources. In this work, we have proposed a novel use of co-occurrence constraints to improve the performance of nonnegative matrix factorization (NMF) ?a popular, convenient, and effective method for decomposing spectrograms. We have developed three new multiplicative update rules to influence the amount of dependence among musical atoms. Here, we demonstrate the ability of these updates to learn more accurate representations of musical objects using multiple atoms. We have also proposed a novel method that imposes additional harmonic constraints upon the musical atoms learned by NMF. When there is significant spectral-temporal overlap among the musical sources, our learning method has better recall and precision than other popular existing matrix factorization methods.
Date: 9/9/2009
Speaker: Yan Chen
Title: Cooperative P2P Streaming: An Evolutionary Game-Theoretic Approach

### Summer 2009

Date: 8/5/2009
Speaker: Matt Stamm
Title: Disguising an Image’s JPEG Compression History Using Anti-Forensic Signal Processing
In recent years, a number of image forensic algorithms have been proposed which rely on detecting JPEG compression artifacts. These algorithms are capable of providing evidence of image manipulations, supplying information about an image’s source, and identifying forged regions within an image. None of these algorithms, however, consider the possibility of an image forger using anti-forensic signal processing operations. In this work, we present an anti-forensic technique for disguising an image's JPEG compression history. We show how the proper addition of noise to an image's discrete cosine transform coefficients can sufficiently remove quantization artifacts which act as indicators of JPEG compression while introducing an acceptable level of distortion. Furthermore, we present a method which removes blocking artifacts while preserving the forensically significant properties of an image’s DCT coefficient distribution. Simulation results are provided to verify the efficacy of this anti-forensic technique.
Date: 7/22/2009
Speaker: Steve Tjoa
Title: MIDI-Informed Music Source Separation using Nonnegative Matrix Factorization
Music source separation has drawn plenty of attention for its ability to facilitate other tasks in music information retrieval. However, source separation is difficult because there does not exist a unique solution; given a mixture, there are several valid ways to decompose the mixture into component sources. Researchers have devoted years of effort toward solving the fully automatic blind source separation problem. Although significant progress has been made along the theoretical and algorithmic aspects of source separation, the separation performance achievable by the state of the art is still not adequate for widespread commercial use. This inadequacy exists in large part because of the prohibitively difficult nature of solving the blind source separation problem. In this talk, we propose a novel method for performing music source separation of a music signal given side information in MIDI format of a similar performance. Our method uses nonnegative matrix factorization -- a popular, convenient, and effective method for decomposing matrices. By exploiting the information that MIDI provides, our method is able to separate sources within highly polyphonic musical mixtures as well as sources which share a significant amount of temporal or spectral overlap.
Date: 7/15/2009
Speaker: Amr El Sherif
Title: Multiple Access Protocols for Cooperative Communications and Networking
In cooperative communications different network nodes share their antennas and resources to form a virtual antenna array and improver their performance through spatial diversity. This thesis contributes to the advancement of cooperative communications by developing and analyzing new multiple access cooperation protocols that leverage the benefits of cooperation to upper network layers.

For speech communications networks, we propose a cooperative multiple access protocol that exploits inherent characteristics of speech signals, namely, long periods of silence, to enable cooperation without incurring bandwidth efficiency losses. Using analytical and simulation results we show that the proposed protocol achieves significant increase in network throughput, reduction in delay, and improved perceptual speech quality.

In TDMA networks, we investigate the problem of sharing idle time slots between a group of cooperative cognitive relays helping primary users, and a group of cognitive secondary users. Analytical results reveal that, despite the apparent competition between relays and secondary users, and even in case of mutual interference between the two groups, both primary and secondary users will significantly benefit in terms of maximum stable throughput from the presence of relays.

For random access networks, we find a solution to the problem of achieving cooperation gains without suffering from increased collision probability due to relay transmissions. A novel cooperation protocol is developed and analyzed for that purpose. Analytical and simulation results reveal significant improvements in terms of throughput and delay performance of the network. Moreover, a decrease in collision probability.

Finally, in the framework of a cognitive radio network, we study the negative effects of spectrum sensing errors on the performance of both primary and secondary networks. To alleviate those negative effects, we propose a novel joint design of the spectrum sensing and channel access mechanisms. Results show significant performance improvement in the maximum stable throughput region of both networks.

Date: 7/1/2009
Speaker: Wan-Yi Lin
Title: Behavior Modeling and Forensics for Multimedia Social Network
Within the past decades, the explosive combination of multimedia signal processing, communications and networking technologies has facilitated the sharing of digital multimedia data and enabled pervasive digital media distribution over all kinds of networks. People involved in the sharing and distribution of multimedia contents form multimedia social networks in which users share and exchange multimedia content, as well as other resources. Users in a multimedia social network have different objectives and influence each other's decision and performance. It is of ample importance to understand how users interact with and respond to each other and analyze the impact of human factors on multimedia systems. This thesis illustrates various aspects of issues and problems in multimedia social networks via two case studies of human behavior in multimedia fingerprinting and peer-to-peer live streaming.

Since media security and content protection is a major issue in current multimedia systems, this thesis first studies the user dynamics of multimedia fingerprinting social networks. Before a collusion being successfully mounted, the colluders must be stimulated to cooperate with each other and all colluders have to agree on the attack strategy. Therefore, not all types of collusion are possible. We reduce the possible collusion set by analyzing the incentives and bargaining behavior among colluders. Furthermore, we investigate the side information which improves the traitor-tracing performance and provide the optimal strategies for both users (fingerprint detector and the colluders) in the multimedia fingerprinting social network. We show that the optimal strategies designed based on human behavior can provide more information to the fingerprint detector and effectively improve the collusion resistance.

The second part of this thesis focuses on understanding modelling and analyzing user dynamics for users in various types of peer-to-peer live streaming social networks. We stimulate user cooperation by designing the optimal, cheat-proof, and attack-resistant strategies for peer-to-peer live streaming social networks over Internet as well as wireless networks. Also, as more and more smart-phone users subscribe to the live-streaming service, a reasonable market price has to be set to prevent the users from reselling the live video. We start from analyzing the equilibrium between the users who want to resell the video and the potential buyers to provide the optimal price for the content owner.

Date: 6/17/2009
Speaker: Yan Chen
Cognitive radio technologies have become a promising approach to efficiently utilize the spectrum. Although many works have been proposed recently in the area of cognitive radio, little effort has been made in content-aware multimedia application over cognitive radio network. In this paper, we study the multimedia streaming problem over cognitive radio network, where there is one primary user and N secondary users. In this problem, the objective of the primary user is to maximize his/her revenue by choosing either to utilize the spectrum himself/herself or to sell the spectrum to the secondary users, while the objective of each secondary user is to maximize his/her payoff by competing with the primary user and other secondary users to buy the spectrum. We formulate the spectrum allocation problem as an auction game and propose to solve the problem using alternative ascending clock auction. We show that the proposed algorithm is cheat-proof and converges to the NE that maximizes social warfare in a finite number of clocks. We prove and demonstrate with simulation results that compared with traditional dual-based optimization algorithm that maximizes social warfare, the proposed method has the cheat-proof property, while compared with second-price sealed-bid auction that has cheat-proof property, the proposed algorithm can achieve much higher social warfare.

### Spring 2009

Date: 5/12/2009
Speaker: Hua Chen
Title: Dynamic Spectrum Sensing under Energy Constraint
In cognitive radio networks, secondary users must sense the primary spectrum bands and only utilize those free spectrum bands where no primary users are present. In order to sufficiently protect the primary user's transmission, the spectrum sensing can take up to 10%~15% of a time slot at low SNRs. In this work, we assume sensing energy constraints on secondary user, i.e. the secondary users can only sense the spectrum for a limited time. We propose to dynamically allocate the spectrum sensing duration to improve the secondary user's payoff functions based on the primary user state. We first discuss the problem under ideal assumption where secondary users have primary user state available at the end of every slot, and then we propose algorithm without this assumption. Simulation results show that our algorithm can improve the secondary user payoff functions by 5%~15%, depending on the assumption we used.
Date: 5/5/2009
Speaker: Matt Stamm
Title: An Anti-Forensic Attack on an Image's JPEG Compression History
The widespread availability of photo editing software has made it easy to create visually convincing digital image forgeries. To address this problem, there has been much recent work in the field of digital image forensics. There has been little work, however, in the field of anti-forensics, which seeks to develop a set of techniques designed to fool current forensic methodologies. In this work, we present a technique for disguising an image's JPEG compression history. An image's compression history can be used to provide evidence of image manipulation, supply information about the camera used to generate an image, and identify forged regions within an image. We show how the proper addition of noise to an image's discrete cosine transform coefficients can sufficiently remove quantization artifacts which act as indicators of JPEG compression while introducing an acceptable level of distortion. Simulation results are provided to verify the efficacy of this anti-forensic technique.
Date: 4/28/2009
Speaker: Yongle Wu
Title: Optimal Power Allocation Strategy against Jamming Attacks Using the Colonel Blotto Game
Cognitive radio technologies have become a promising approach to increase the efficiency of spectrum utilization. Although cognitive radio has been intensively studied in recent years, only a few works have discussed security aspects. In this paper, we focus on the jamming attack, one of major threats to cognitive radio networks, where a malicious user wants to jam the communications of secondary users by injecting interference. Aware of the absence of several primary users and the presence of a malicious user, a secondary user can allocate power to those fallow bands with a randomized strategy, in hope of alleviating the damage caused by the malicious user. We model this scenario into a two-player zero-sum game, and derive its unique Nash Equilibrium under certain conditions using the Colonel Blotto game approach, which provides a minimax strategy that the secondary user should adopt in order to minimize the worst-case damage caused by the malicious user. Simulation results are presented to verify the performance.
Date: 4/14/2009
Speaker: Amr El Sherif
Title: Cooperation in Random Access Networks
In this work our goal is to answer the questions of how to enable cooperative communications in random access networks? And, since cooperation introduces extra transmissions in the channel, what are the benefits and possible tradeoffs associated with cooperation? To achieve this goal, we first propose a cooperation protocol that enables relays to exploit part of the channel resources to offer diversity. Next, a Markov model that characterizes the interaction between relays and different network nodes is developed to investigate the effect of sharing the channel resources between network nodes and relays on the throughput and collision probability. Analytical and numerical results reveal significant performance increase by the virtue of our cooperative protocol.
Date: 4/7/2009
Speaker: Ahmed Ibrahim
Title: Relay Deployment and Selection in Cooperative Wireless Networks
In cooperative communication protocols, multiple terminals cooperate together forming a virtual antenna array to improve their performance. This thesis contributes to the advancement of cooperative communications by proposing new relay deployment and selection protocols across the network layers that can increase the bandwidth efficiency, reduce the end-to-end transmission power needed to achieve a desired network throughput, maximize the lifetime of a given network, rebuild a disconnected network, and mitigate the effect of channel estimation error and co-channel interference (CCI) problems.

Conventional cooperative schemes achieve full diversity order with low bandwidth efficiency. In this thesis we propose a relay selection cooperative protocol, which achieves higher bandwidth efficiency while guaranteeing full diversity order. We provide answers to two main questions, namely, "When to cooperate?" and "Whom to cooperate with?". Moreover, we obtain optimal power allocation and present the tradeoff between the achievable bandwidth efficiency and the corresponding symbol error rate performance.

We illustrate that the cooperation gains can be leveraged to the network layer. In particular, we propose a cooperation-based routing algorithm, namely, the Minimum Power Cooperative Routing (MPCR) algorithm, which optimally selects relays while constructing the minimum-power route. Moreover, the MPCR can be implemented in a distributed manner. Using analytical and simulation results, we show that the MPCR algorithm achieves significant power savings compared to the current cooperation-based routing algorithms.

We also consider maximizing the network lifetime in sensor networks via deployment of relays. First, we propose a network maintenance algorithm that obtains the best locations for a given set of relays. Second we propose a routing algorithm, namely, Weighted Minimum Power Routing algorithm, that significantly increases the network lifetime due to the efficient utilization of the deployed relays. Finally, we propose an iterative network repair algorithm that finds the minimum number of relays along with their best locations, needed to reconnect a disconnected network.

We complete this thesis by investigating the impact of cooperative communications on mitigating the effect of channel estimation error and CCI. We show that cooperative transmission schemes are less susceptible to the effect of channel estimation error or CCI compared to the direct transmission. Finally we study the tradeoff between the timing synchronization error, emerging in the case of having simultaneous transmissions of the cooperating relays, and the channel estimation error, and show their net impact on the system performance.

Date: 3/24/2009
Speaker: Wan-Yi Lin
Title: Game-Theoretic Analysis of Colluder Social Networks
Users in multimedia social networks actively interact with each other. It is crucial to study the complex user dynamics and analyze its impact on the performance of multimedia social networks. This paper provides a case study of user dynamics in multiuser collusion attacks against multimedia fingerprinting. During collusion, a group of attackers collectively and effectively attack multimedia fingerprinting system and use multimedia content illegally. We present a game-theoretic framework to model the complex dynamics among colluders, analyze when attackers cooperate with each other, and investigate how a colluder selects his/her fellow attackers to maximize his/her own payoff. Our framework considers both the colluders?risk of being detected by the digital rights enforcer and the reward received from illegal usage of multimedia content. Our analysis shows that colluding with more attackers does not always increase an attacker’s utility, and attackers may not always want to cooperate with each other. We first examine the necessary conditions for attackers to collude together, and study how they select the collusion parameters such that cooperation benefits all colluders. We further study the bargaining process and the fairness solutions for collusion including the time-sensitiveness of multimedia contents. We then study how the number of colluders affects each attacker’s utility, and investigate the optimum strategy that an attacker should use to select fellow attackers in order to maximize his or her own payoff.
Date: 3/3/2009
Speaker: Steve Tjoa
Title: Harmonic Dictionary Learning for Music Source Separation using Nonnegative Singular Value Decomposition
Music source separation, an important task in the area of music information retrieval (MIR), has received plenty of attention in the past few years for its potential to improve the performance of other MIR-related tasks. While related to other generic problems in blind source separation, the musical context of this problem contains additional challenges which diminishes the separating power of common solutions, such as independent component analysis (ICA). At the same time, the musical context also contains additional cues which can improve separation performance if exploited properly. In this work, we investigate matrix factorization methods, particularly the nonnegative singular value decomposition (NNSVD), to factorize a time-frequency representation of a musical mixture into its component source signals. Furthermore, we offer a novel contribution to the NNSVD learning process which further exploits the harmonic nature of the input music signal. We demonstrate the effectiveness of this approach through the separation of synthetic and natural music signals, and we compare the approach against other traditional matrix factorization methods.
Date: 2/24/2009
Speaker: Mohammed W. Baidas
Title: Energy Efficiency of Cognitive Cooperative Distributed Detection in Wireless Sensor Networks
Cooperative communications have been proposed to exploit spatial diversity gains. In this paper, binary hypothesis distributed detection problem in wireless sensor networks with cognitive-cooperative relay deployment is considered. In energy-constrained networks such as wireless sensor networks, the advantages of cooperation can be further exploited by cognitively utilizing wasted channel/time slots among sensor nodes for optimal data gathering and detection. Wireless sensor networks are also characterized by correlated observations and sending all observations directly to the fusion center all the time is a waste of network-resources. By employing cognitive cooperative relays significant detection performance gains and energy savings can be achieved. A simple example of a binary hypothesis distributed detection problem in a spatio-temporally correlated sensor network with two sensor nodes and a relay node illustrates the significant energy savings achieved through cognitive-cooperation.
Date: 2/17/2009
Speaker: Beibei Wang
Title: Game-Theoretic Dynamic Spectrum Allocation over Cognitive Radio Networks
Wireless spectrum is a very precious resource. However, a large portion of the assigned spectrum is used sporadically. The limited spectrum resources and the inefficient spectrum usage necessitate a new communication paradigm to exploit the existing wireless spectrum opportunistically, i.e., dynamic spectrum access and cognitive radio networks. With cognitive radio technology, future wireless devices are envisioned to be able to utilize the spectrum in a dynamic manner. However, in de-centralized cognitive radio networks, selfish users only aim at maximizing their own benefits. In addition, they are sometimes uncertain about the surrounding environments, such as who else are in the network and what the others do. Thus, it is very important to analyze users?intelligent cooperation and competition behavior from a game-theoretic perspective, as game theory is a well-developed mathematical tool that studies the strategic interactions among rational decision makers under uncertainty. My talk today focuses on how to solve dynamic spectrum allocation problems using game-theoretic approaches. The first part is on relay selection and power control using Stackelberg game, and the second part is about collaboration strategy in cooperative spectrum sensing using evolutionary game modeling.

### Fall 2008

Date: 11/25/2008
Speaker: Hua Chen
Title: A survey of Cognitive Radio from Information Theory perspective
Recently cognitive radio has attracted intense research interests as an enabling technology for dynamic spectrum access. The research on cognitive radio involves interdisciplinary effort from various technical aspects, including signal processing, communications, cooperation/game theory. However, there are still a lot of open problems about cognitive radio from perspective of information theory. In this talk, we will give an overview of the challenges faced by information theory research of cognitive radio, and discuss some potential research topics for future work.
Date: 11/04/2008
Speaker: Amr El Sherif
Title: Sensing Errors Aware Dynamic Spectrum Access Strategy
In a dynamic spectrum access system, secondary users try to select their different sensing and transmission parameters to achieve a balance between two conficting goals. The first is to make their activity transmarent to the primary users, while the second goal is maximizing their own throughput. Due to the inevitable occurance of spectrum sensing errors, collisions between primary and secondary transmissions will occur and lead to decreased throughput for both primary and secondary users. In this work, we study the effect of spectrum sensing erros on the system's performance from a stable throughput point of view, and we propose a novel spectrum access strategy that is aware of the spectrum sensing errors and that aims at maximizing the stability region for both primary and secondary users. The design of the access stratgey is formulated as an optimization problem, and different approximations are proposed to simplify the solution of this problem are developed. Our results show that the exact and approximate solutions are very close to each other, and that the proposed strategy offers a significant incrase in the throughput of both primary and secondary users.
Date: 10/28/2008
Speaker: Matt Stamm
Title: Lung Tumor Motion Compensation Using a Robotic Treatment Couch: Problem Overview
Radiotherapy is a common form of tumor treatment in which a patient is subjected to localized radiation from a particle beam aimed directly at a cancerous tumor. To avoid unnecessary damage to healthy tissue, care must be taken to immobilize the tumor during treatment. For tumors of the lung, this is not easily possible due to tumor motion resulting from breathing. In order to mitigate this effect, a robotic couch with feedback control has been proposed to compensate for a patient's breathing signal. Past work on this project is described and future work is outlined.
Date: 10/21/2008
Speaker: Yan Chen
Title: Multi-User Rate Allocation Game For Multimedia Communications
How to efficiently and fairly allocate data rate among different users is a key problem in the field of multiuser multimedia communication. However, most of the existing optimization-based methods, such as minimizing the weighted sum of the distortions or maximizing the weighted sum of the PSNRs, have their weights heuristically determined. Moreover, those approaches mainly focus on the efficiency issue while ignoring the more important fairness issue. In this paper, we address this problem by proposing a game-theoretic framework, in which the utility/payoff function of each user/player is jointly determined by the characteristics of the transmitted video sequence and the allocated bit-rate. We show that with the proportional fairness criterion, the game has a unique Nash equilibrium, according to which the controller can efficiently and fairly allocate the available network bandwidth to the users. We also show that the traditional optimization-based approach is a special case of the game-theoretic framework with ill-defined utility function. Finally, we show several experimental results on real video data to demonstrate the efficiency and effectiveness of the proposed method.
Date: 10/07/2008
Speaker: Beibei Wang
Title: Evolutionary Game Framework for Cooperative Spectrum Sensing with Selfish Users: How to Cooperate?
Cooperative spectrum sensing has been shown to be able to greatly improve the sensing performance in cognitive radio networks. However, all secondary users cooperating in sensing at every time slot may not be optimal, or even difficult if the cognitive users belong to different service providers, where they tend to contribute less in sensing in order to increase their own throughput. In this paper, we propose an evolutionary game framework to study the time evolution of cooperation of the selfish users in spectrum sensing. We derive the behavior dynamics and the evolutionarily stable strategy (ESS) of the secondary users by using replicator dynamics. We then prove that the dynamics converge to the ESS, which renders the possibility of a decentralized implementation of the proposed sensing game. According to the dynamics, we further develop a distributed learning algorithm so that the secondary users approach the ESS only with their own payoff observation. We finally study the possible attacks on the spectrum sensing game and analyze their damages. Simulation results show that the average throughput achieved in the cooperative sensing game is higher than that when the secondary users sense the primary user individually without cooperation. The proposed game is demonstrated to converge to the ESS, and achieve a better system performance than the scheme having all user always contribute to sensing.
Date: 09/30/2008
Speaker: Steve Tjoa
Title: Musical Source Separation using Harmonic Clustering
The task of separating music into its constituent sources has been a central focus of music information retrieval for many years. Aside from the aesthetic reasons for separating music (e.g., remixing audio), source separation simplifies many other problem domains in music information retrieval such as transcription, audio fingerprinting, cover song retrieval, rhythm representation, and more. Unlike blind source separation in general signal scenarios, musical settings impose further constraints which render these generic solutions obsolete. In this talk, we introduce the use of clustering in the harmonic feature vector domain by using pitch labels as priors. This approach increases the coherency within each separated source over other approaches which do not perform clustering in this domain.
Date: 09/23/2008
Speaker: Yongle Wu
Title: A Scalable Auction Mechanism
Dynamic spectrum access (DSA), enabled by cognitive radio technologies, has become a promising approach to improve efficiency in spectrum utilization, and the spectrum auction is one important DSA approach, in which secondary users lease some unused bands from primary users. In this paper, we propose a novel multi-winner spectrum auction game to achieve optimal spectrum allocation and high seller's revenue. As secondary users may be selfish in nature and tend to be dishonest in pursuit of higher profits, we develop effective mechanisms to suppress their dishonest/collusive behaviors when secondary users distort their valuations about spectrum resources and interference relationships. Moreover, in order to make the proposed game scalable when the size of problem grows, the semi-definite programming (SDP) relaxation is applied to reduce the complexity significantly. Finally, simulation results are presented to evaluate the proposed auction mechanisms, and demonstrate the complexity reduction as well.
Date: 09/16/2008
Speaker: Quoc Lai
Title: Wireless Network Cocast
In this talk, a new communication paradigm, namely "Wireless Network Cocast (WNC)", is proposed. WNC utilizes cooperative communication and linear network coding to achieve desired diversity order with low transmission delay and low cost. In particular, for a network of N nodes, a diversity order N is achieved with only 2N time slots, a significant improvement in comparison with a conventional cooperative communication scheme, namely maximal cooperation (MAX), which requires NxN time slots for diversity order of N. Talk starts with defining the problem with direct transmission (DTX). Then current solutions using relay tower and cooperative communication are presented and their shortages are discussed. The discussion provides the motivation for the new approach WNC. The transmission structure of WNC with linear network coding is then presented, the performance is analyzed, and simulation reveals its advantageous performance. A number of WNC applications are also presented. In particular, location-aware WNC and WNC with relay tower will be discussed. At the end, the talk is concluded by providing the advantages of WNC.
Date: 09/09/2008
Speaker: Wan-Yi Lin
Digital fingerprinting is an emerging technology that offers proactive post-delivery protection of multimedia. Multiuser collusion attack is a powerful attack against digital fingerprinting, where a group of attackers collectively mount attacks to attenuate the identifying marks. When the collusion-resistant forensic code is applied onto multimedia content, the colluders usually perform post-processing on the colluded multimedia content before redistribution. The post-processing incurs error in the detected bits of the forensic code and posts a new challenge to the forensic code design. We designed the anti-collusion and post-processing resistant binary forensic by concatenating orthogonal codes with low correlation Reed-Solomon code. The simulation results shows that our forensic code design can achieve perfect detection when the number of colluders is less than 5 and up to 45% of the forensic code is changed for a five-minute video clip.

### Summer 2008

Date: 08/06/2008
Speaker: Hua Chen
Title: RF Time-Reversal Architecture
Traditional wireless communication techniques cannot handle severe multipath and scattering as they cannot know precisely the channel impulse response. In this talk, we present a new technique - RF time-reversal architecture (RF-TRA), in which the RX sends an impulse into the system and hence the TX obtains perfect channel impulse response, provided the channel is reciprocal. Based on the channel response, the TX sends a reconstructed signal which automatically focuses power back to the RX. The implementation of RF-TRA system relies on high speed sampling ADCs/DACs and high speed signal processing on FPGA. We also show that it is possible to build up a preliminary demo at a lower data throughput using the Universal Software Radio Peripheral (USRP) board and GNU Radio.
Date: 07/30/2008
Speaker: Matt Stamm
Title: Blind Detection of Additive Noise Tampering for Digital Image Forensics
Digital images, as well as digital image editing software, have become pervasive in today’s society. As a result, it has become necessary to develop a family of tests to determine if a digital image has been manipulated, or if it exists in an unaltered state. One forensically important image alteration is the insertion of additive noise into an image. While this operation itself may not alter the perceptual content of an image, it can be used to effectively mask traces of other tampering operations. In this talk we present a blind method to detect the insertion of additive noise into a digital image. This method operates by determining the presence and strength of statistical artifacts which arise in an image due to the insertion of additive noise. Simulation results for the proposed detection scheme indicate that a probability of detection in excess of 95% can be achieved for a probability of false alarm below 3%.
Date: 07/23/2008
Speaker: Amr El Sherif
Title: Vehicular Ad Hoc Networks: An Overview
Vehicular Networks are an emerging type of networks in which vehicles and roadside units are the communicating nodes; providing each other with information, such as safety warnings and traffic information. As a cooperative approach, vehicular communication systems can be more effective in avoiding accidents and traffic congestions than if each vehicle tries to solve these problems individually. In this talk, we will give a general overview of vehicular networks and there applications, the enabling technologies and will identify some interesting research issues for future work.
Date: 07/16/2008
Speaker: Yongle Wu
Title: Cognitive Radio Networks under Malicious Attacks
In this talk, we investigate impact of jamming attacks on cognitive radio networks, a very important yet rarely discussed problem. We assume a secondary network coordinated by a secondary base station is attacked by an energy-limited jamming attacker. The M/G/1 queue with vacation model is employed to provide quantitative analysis on damages in terms of average waiting time and blocking probability. When the traditional collaborative sensing scheme is used, malicious attackers can easily capture the slot pattern by monitoring the spectrum. Then, they can launch efficient attacks by jamming the sensing phase only. In order to make the cognitive radio networks more robust to malicious attacks, we propose a frequency-hopping collaborative sensing scheme to alleviate the damages caused by energy-limited attackers.
Date: 07/09/2008
Speaker: Yan Chen
Title: Risk Distortion Analysis For Video Collusion Attack
In this talk, we investigate the relationship between risk and distortion for the linear video collusion attack with Gaussian fingerprint. By modelling the residue, i.e. the different between temporal adjacent frames, as a Gaussian distribution, we express the risk and distortion as functions of the temporal filter coefficients. By adjusting the coefficients, we are able to find the minimal distortion under a pre-defined risk constraint. We show that, given a fixed false alarm probability \alpha, when the risk is not larger than \alpha, the globally optimal coefficient can be found by solving a convex optimization problem using numerical method, e.g. interior points method. When the risk is larger than \alpha, the problem of finding the globally optimal coefficient is not convex (with a convex objective function but concave-convex constraints). However, the locally optimal coefficient can be found by the constrained concave-convex procedure (CCCP). Using the optimal coefficient (either globally or locally), an implicit risk-distortion model can be obtained. Furthermore, we derive an explicit risk-distortion mode by assuming the variance of the residue (a quadratic form of the coefficients) is around a constant, e.g. \beta. We also discuss the optimal detection and attack with side information. Finally, we conduct several experiments to verify the proposed risk-distortion model using real video data.
Date: 07/02/2008
Speaker: Beibei Wang
Title: Evolutionary Game Framework for Behavior Dynamics in Cooperative Spectrum Sensing
Cooperative spectrum sensing has been shown to greatly improve the sensing performance in cognitive radio networks. However, if the cognitive users belong to different service providers, they tend to contribute less in sensing in order to achieve a higher throughput. In this paper, we propose an evolutionary game framework to study the interactions between selfish users in cooperative sensing. We derive the behavior dynamics and the stationary strategy of the secondary users, and further propose a distributed learning algorithm that helps the secondary users approach the Nash equilibrium with only local payoff observation. Simulation results show that the average throughput achieved in the cooperative sensing game with more than two secondary users is higher than that when the secondary users sense the primary user individually without cooperation.
Date: 06/25/2008
Speaker: Steve Tjoa
Title: Supervised Classification of Musical Content by Genre and Artist
With the explosion of digital music distribution and sharing over the past decade, the growth of huge musical databases have created a need for fast and accurate information retrieval and classification of music. However, due to a lack of consensus among music listeners over the definition of various genres, automatic classification remains a difficult task. Today, we will discuss methods used for classification of musical genres and artists using supervised methods along with the correspondence between feature extraction and musical properties such as timbre, rhythm, and harmony. The results of our feature extraction will be shown in conjunction with supervised classification methods such as support vector machines, linear discriminant analysis, and K-nearest neighbor search.
Date: 06/18/2008
Speaker: Mohammed Baidas
Title: Correlation-Based Cooperation for Distributed Detection in Wireless Sensor Networks: Analysis and Modeling
A detection problem in sensor networks is considered where sensor nodes are modeled on two concentric circles and receive partial information about their environment. The nodes transmit their observations over a noisy channel to the fusion center for the purpose of detection. In this work, we study the system performance as a function of sensor density which allows us to explore the natural tradeoff between the total number of sensor nodes and the quality of information. In dense sensor networks with spatially correlated sensors, it is likely that sensors within a close vicinity of each other would observe highly correlated data. While conditional independence is a convenient and widely used assumption, it is likely to fail in such a dense network. So, instead of competing for resources and causing network congestion, highly correlated sensors should cooperate and share the use of transmission channels for optimal data gathering. Our analysis is evaluated in terms of the probability of detection error for different simple scenarios in order to characterize the effect of inter-sensor separation into the selection of sensors/relays for cooperation.

### Spring 2008

Date: 05/09/2008
Speaker: Matt Stamm
Title: A New Blind Method for Detecting Digital Image Forgeries
The development of powerful image editing software such as Adobe Photoshop has made it possible to construct inauthentic images that are visually realistic. In response, a need has arisen for the development of techniques which can identify tampering in digital images. Digital watermarking technologies are often ill suited to address this problem because they require insertion of the watermark before tampering has occurred. Other existing tamper identification techniques require either knowledge of the device used to generate the image, or a large training dataset of images from the same source. In this presentation, we propose a new method which detects image forgeries using no a priori knowledge of the image or digital capture device. This method operates by detecting statistical artifacts left behind by nonlinear pixel mappings involved in creating the digital forgery. A full description of the detection algorithm, as well as simulation results are presented in this talk.
Date: 05/02/2008
Speaker: Yongle Wu
Title: Collusion-Resistant Multi-Winner Spectrum Auction for Cognitive Radio Networks
In order to fully utilize spectrum, auction-based dynamic spectrum allocation has become a promising approach which allows unlicensed wireless users to lease unused bands from spectrum license holders. Because spectrum resources are reusable by users far apart, in some scenarios, spectrum is more efficiently utilized by awarding one band to multiple secondary users simultaneously, which distinguishes it from traditional auctions where only one user can be the winner. However, the multi-winner auction is a new concept posing new challenges in the traditional auction mechanisms, because such mechanisms may yield low revenue and are not robust to some newly-emerging collusion. Therefore, in this paper, we propose an efficient mechanism for the multi-winner spectrum auction with collusion-resistant pricing strategies, in which the optimal spectrum allocation can be solved by binary linear programming and the pricing is formulated as a convex optimization problem. Furthermore, a greedy algorithm is proposed to reduce complexity for multi-band auctions. Simulation results are presented to evaluate our proposed auction mechanisms.
Date: 04/25/2008
Speaker: Amr El Sherif
Title: Joint Design of Spectrum Sensing and Access Protocols
The basic idea behind oppotunistic spectrum access is to allow secondary users to search and exploit instanteneous spectrum opportunities while limiting the level of iinterference perceived by primary users. Due to noise and fading, spectrum sensing errors are inevitable. These errors results in increased interference to primary users and decreased throughput for secondary users. Therefore, interaction between physical layer sensing and MAC layer access starategies always exist. In this work, we study this interaction and propose a joint design of spectrum sensing and channel access protocols. Our design is analyzed from a queuing theory point of view, and numerical results are used to demonstrate significant throughput improvements using our joint design method.
Date: 04/18/2008
Speaker: Karim Seddik
Title: Diversity in Cooperative Networks: How to achieve and Where to Exploit?
Date: 03/28/2008
Speaker: Quoc Lai
Title: Cooperative Transmission and Information Aggregation for Wireless Sensor Networks
In wireless sensor networks, reducing total transmit power, required by all sensors to send their information to the sink, is one of the key design issues. In addition, having an even distribution of transmit power among sensors increases network lifetime. Typically, the information gathered from different sensors is spatially correlated to each other. Conventional information aggregation schemes, making use of this information correlation, require high total transmit power and achieve unfair power distribution. In this paper, we propose cooperation-based information aggregation schemes that utilize cooperative communication to reduce the transmit power and achieve even power distribution. We show that by utilizing cooperative communication among sensors, our proposed schemes yield more than 67% in power reduction for any network sizes. For network sizes greater than 5 sensors, the power reduction is more than 90%. In addition, the transmit power is more uniformly distributed in our proposed schemes.

Title: Location-Aware Cooperative Communications utilizing Linear Network Coding
Cooperative communication can be used to reduce the transmit power of distant mobile units, compared to conventional direct transmission, given the same quality-of-service. However, imposing the constraint of having orthogonal transmission for the source and relays leads to large delay in TDMA systems. For a network of $N$ mobile units, the transmission delay would be N(N+1)/2. In this work, we propose a location-aware cooperation-based scheme that aims to reduce transmit power of distant mobile units while maintaining a low transmission delay. The scheme utilizes a linear network coding protocol, where each mobile unit applies linear network coding to a set of transmit symbols that it has received previously. At the base station, multiuser detection is used to decouple the transmit symbols. Both decode-and-forward and amplify-and-forward cooperative protocols are considered. We show that our proposed scheme achieves a comparable bit-error-rate performance with the conventional cooperation-based TDMA scheme while requiring a delay of (2N-1) time slots, a substantial reduction in the transmission delay.
Date: 03/14/2008
Speaker: Steve Tjoa
Title: Strategies for Detecting Digital Audio Forgeries at the Signal Level
The prevalence of digital audio editors in today's society allow us to conveniently and efficiently record and edit music. Unfortunately, these editors also enable malicious users to create digital audio forgeries more easily than ever before. In recent cases of audio forgery, attackers use an existing audio file, perform little or no modification upon the file, and then call the file as their own. In this talk, we propose a novel and effective detection strategy which allows us to accurately match forged audio signals to their original sources. Aside from forgery detection strategies, our contributions include the examination of attacks at the signal level, use of attack-invariant feature extraction, performance improvements via pattern classification methods, and comparison against existing audio fingerprinting systems.
Date: 03/07/2008
Speaker: Ahmed Ibrahim
Title: Self-Noise Mitigation via Cooperative Communications
In this work, we investigate the impact of the cooperative communications on the self-noise problem in cellular networks. The self-noise phenomena expresses the portion of the transmission power that contributes to the noise due to channel estimation errors. First, we derive the outage probability for the direct and cooperative transmission scenarios taking into consideration the self-noise problem. We show that the outage probability reduces by utilizing cooperative communication protocols. Second, we derive the signal-to-noise ratio (SNR) gap ratio, which measures the relative SNR gap due to the self-noise problem. We show that the decode-and-forward cooperative scheme is less susceptible to the channel estimation errors compared to the direct transmission one. Finally, we investigate the effect of the self-noise problem on relay-selection cooperative and multi-path direct transmission schemes.
Date: 02/29/2008
Speaker: Beibei Wang
Title: Dynamic Frequency-Intelligent Reserve-and-Switch Technique (D-FIRST) to Combat Inter-Operator Interference
In this paper, a spectrum sharing scheme that will coordinate among different co-existing cellular operators competing for the same spectrum band is proposed. Based on this scheme, the cell of an operator can be divided into several sub-regions, and mobile stations (MSs) inside each sub-region form one subset. The whole frequency band assigned to a cell is partitioned into slots dedicated to the subsets based on the Quality of Service (QoS) demand. When interference from other operators is detected, the victim operator can switch the frequency of the interfered MSs with the MSs in the safe region, and/or switch to the reserved band. In this way, the inter-operator interference (IOI) can be reduced. From the simulation results, it is shown that with the proposed protocol, the total power consumption of both operators can be reduced significantly. Furthermore, it has been demonstrated that in order to reduce the IOI in a high-density area, the operator should reserve more bandwidth for potential frequency-switching.
Date: 02/08/2008
Speaker: Wan-Yi Lin
Title: Multimedia Social Network Analysis:Incentive-based Secured Peer-to-Peer Live Streaming
Multimedia social network analysis is a research area with rising importance, which analyzes the behavior of users who share multimedia content and investigates the impact of human dynamics on multimedia systems. Members in peer-topeer live-streaming social networks cooperate with each other to provide a distributed, highly scalable and robust platform for live streaming applications. Since every member has different upload/download bandwidth, to accommodate better service, scalable video coding should be used, which challenges best chunk-request strategy. We propose a cheat-proof optimal strategy for 2 user case and extend it to multiple user scenario.

### Fall 2007

Date: 11/21/2007
Speaker: Amr El Sherif
Title: Throughput and Delay Analysis in Cognitive and Cooperative Networks
Cooperative communications and spectrum sharing, originally considered as separate problems, are in fact dual problems in the sense that he available unused or under-used channel resources can be utilized to enhance the primary system performance via cooperation, or it can e shared by a secondary system to transmit new information. In this work, we propose a multiple access protocol that enables both relays helping primary users, and secondary to coexeist and share the free available resources in the network. We provide analytical and simulations results for the throughput and delay performance of the proposed protocol.
Date: 11/14/2007
Speaker: Mahmoud Abdulrehem
Title: Distributed Spectrum Sensing for Opportunistic Spectrum Access
Cognitive radio is a candidate technology for more efficient spectrum utilization systems based on opportunistic spectrum sharing. Because this technology does not rely on traditional license-based spectrum allocation policies, it could disrupt existing systems if the spectrum utilization decision is based on unreliable spectral estimation. Today, we are going to investigate the concept of cooperative spectrum sensing in fading channel based on distributed spectrum sensing. We also considered the security problem in cooperative spectrum sensing. We proposed a new algorithm to identify the presence of attackers and eliminate their effect on the system.
Date: 11/07/2007
Speaker: Matt Stamm
Title: Blind Detection of Contrast Enhancement in Digital Images
Due to the prevalence of digital cameras and photo editing software, it has become increasingly difficult to determine if a digital image has been altered. Often, alterations to a digital image will leave behind statistical artifacts which can be detected by a third party. In this talk I will propose a blind detection scheme to determine if a global contrast enhancement operation has been used to alter a digital image. In addition a separate blind detection will be proposed for a specific enhancement operation known as histogram equalization. The design procedure and performance of both detection schemes will be discussed.
Date: 10/31/2007
Speaker: Steve Tjoa
Title: Digital Audio Forgery and the Signal Level: Attack Formulation and Detection
Recent cases of audio forgery have been uncovered within the past year, taking the audio community by surprise. In these types of cases, the attacker uses an existing audio file, performs little to no modification upon the file, and then calls the file as his or her own. In this talk, we describe more sophisticated attacks on digital audio at the signal level that result in a large distance between the original and attacked audio, thus avoiding basic quantitative matching methods. Canonical parameterization of sinusoids -- phase, frequency, and amplitude -- are investigated, along with other attacks. Then we offer detection strategies based on audio registration which can perform where quantitative matching or audio fingerprinting/hashing fails.
Date: 10/24/2007
Speaker: Ahmed Ibrahim
Title: Connectivity-Aware Network Maintenance and Repair via Relays Deployment
In this paper, we address the network maintenance problem where we aim to maximize the lifetime of a sensor network by adding a set of relays to it. The network lifetime is defined as the time until the network becomes disconnected. The Fiedler value, which is the algebraic connectivity of a graph, is used as an indicator of the network health. First, we present a network maintenance algorithm that obtains the best locations for a given set of relays. This algorithm depends on formulating the search problem as a standard semi-definite programming (SDP) that can be solved efficiently. Second we propose a routing algorithm, namely, Weighted Minimum Power Routing (WMPR) algorithm, that significantly increases the network lifetime due to the efficient utilization of the deployed relays. Third, we propose an adaptive network maintenance algorithm that relocates the deployed relays based on the network health indicator. We show that the proposed algorithms achieve network lifetime gain of 449.5% by adding 4 relays to a sensor network of 20 nodes. Finally, we consider the network repair problem, in which we find the minimum number of relays along with their best locations to reconnect a disconnected network. We propose an iterative network repair algorithm that utilizes the network maintenance algorithm.
Date: 10/17/2007
Speaker: Karim Seddik
Title: Distributed Detection in Wireless Sensor Networks
In this talk, the problem of distributed detection in wireless sensor networks is considered. A system consisting of a set of sensor nodes communicating to a fusion center, where decisions are made, is considered. We consider the problem of employing relay nodes in the sensor networks by assigning some of the system resources for relaying the information from some of the sensor nodes. As some sensor nodes provide less-informative" measurements to the fusion center, we consider reassigning the system resources allocated for these sensors to relay nodes to forward the measurements of the more-informative" sensor nodes. There will be a trade-off between the number of measurements sent to the fusion center and the reliability of the more-informative measurements. We will analyze the performance of two protocols, namely, Protocol I and Protocol II. In Protocol I, each sensor node directly transmits its measurement to the fusion center. In Protocol II, some of the sensor nodes relay the measurements of other more-informative sensor nodes. Hence, in Protocol II, the reliability of the more-informative measurements is enhanced at the expense of having less measurements sent to the fusion center and this creates the trade-off between the number of measurements and the reliability of the measurements. We analyze the performance of the two protocols over additive white Gaussian noise (AWGN) and Rayleigh fading channels. Based on the analysis, we characterize the regions where the performance of one protocol is superior the other. The results show that in some cases it is better to allocate some of the system resources to relay nodes, not to sensor nodes, to increase the reliability of the more-informative measurements and this leads to a better overall detection performance at the fusion center.
Date: 10/3/2007
Speaker: Beibei Wang
Title: A Correlated Equilibrium Framework for Distributed Resource Allocation Over Wireless Networks
Game theoretical approaches have been shown powerful to achieve distributed resource allocation in wireless networks. However, most of them are based on the concept of Nash Equilibrium (NE), which is generally not system efficiency. Therefore, in this paper we develop a correlated equilibrium (CE) framework for distributed resource allocation in a variety of wireless networking scenarios, such as multiple access control, sensor/ad hoc networks, and dynamic spectrum access networks. Using the CE strategy, wireless users can cooperate on the joint distribution of the actions so as to achieve mutual benefits, which is much better than that of NE strategies. We target to answer two questions: how to utilize the CE concept to model the resource allocation problems, and how to achieve the optimal CE. We first investigate and analyze the properties of the CE convex hull. Then, no-regret learning algorithm is shown, and we further propose a heuristic algorithm allowing limited information exchange among wireless mobile users. From the simulation results, we show that compared with the no-regret learning algorithm, the proposed heuristic learning algorithm not only can achieve better aggregated expected utility, but also converge with a higher speed with small extra signaling overhead.
Date: 9/26/2007
Speaker: Yongle Wu
Title: Repeated Spectrum Sharing Game
The unlicensed spectrum sharing problem is modeled into game theoretic framework. By using the punishment based repeated game, wireless users can automatically maintain some kinds of cooperations, resulting in an more efficient usage of the spectrum. Several cooperation rules are proposed. Furthermore, we propose a mechanism design method to give the users the incentive to reveal their private information truthfully all the time.
Date: 9/12/2007
Speaker: Wan-Yi Lin
Title: Communication Forensics—Modulation Identification over Wireless Network
Modulation forensics is to detect the modulation type in wireless communications without any prior information. It finds both military and civilian applications such as surveillance and cognitive radio. It is a challenging task, especially in a non-cooperative environment, as no prior information on the incoming signal is available at the receiver. In this paper, we investigate the modulation forensics of linear digital modulations and space-time orthogonal code in slowly varying frequency-selective fading channels. With unknown channel vector, and phase distortion at the receive-side, we derive a composite test consisting second-moment nonlinearity and maximum likelihood test, and discuss the performance and forensic system confidence measure. It is shown that the proposed algorithm achieves almost perfect identification of the space-time coding, and high accuracy rate of modulation type detection.
Date: 9/5/2007
Speaker: Quoc Lai
This tutorial will address Multiple Input Multiple Output (MIMO) and related emerging Cooperative Communications technologies and their potential military applications. Both MIMO and Cooperative Communications have the potential to provide enhanced performance capabilities in challenging RF environments such as urban, subterranean and any environment where multi path is a factor. This tutorial will provide an overview of these technologies and address the challenges of incorporating them in a military environment to include mobility issues, size, weight and power, and other factors unique to the military environment. The tutorial will address areas that US Army CERDEC has an interest in pursuing further research and transition of the technology to future communications platforms.

### Summer 2007

Date: 8/8/2007
Speaker: Steve Tjoa
Title: Conditions for Detecting Multiple Instances of Signal Compression
In this work, we revisit the forensic problem of source coder identification for digital images. We extend our earlier work which offered analysis and detection of intrinsic fingerprints generated by lossy image source coders. The new aspect to consider is the presence of multiple instances of compression in a digital images. Such a scenario is common in many typical image processing datapaths. By analyzing quantizers of various complexity applied to various signals, we offer conditions where multiple compression becomes detectable, as well as additional measures one can perform to improve the identifiability of multiple compression.
Date: 8/1/2007
Speaker: Mahmoud Abdulrehem
Title: Modified Primary-Prioritized Markov Approach for Cognitive Radio Spectrum Allocation
Cognitive radio is a candidate technology for more efficient spectrum utilization systems based on opportunistic spectrum sharing. In a dynamically changing spectrum environment, it is very important to consider the statistics of different users' spectrum access so as to achieve more efficient spectrum allocation.

In this work, we propose a modification for the primary-prioritized Markov approach for dynamic spectrum access which is based on modeling the interactions between the primary and the secondary users as continuous-time Markov chains (CTMC). By optimizing the access probabilities for each secondary user, we can improve the performance more than the original proposed algorithm.

Title: A Testbed for Cognitive Radio
In the second talk, we are going to introduce our testbed used for cognitive radio algorithms implementation. This testbed consists of 2 Universal Software Radio Peripheral (USRP) boards and a PC running GNURadio on Ubuntu Linux. We will demonstrate two experiments we have recently done.
Date: 7/18/2007
Speaker: Amr El Sherif
Title: Cognitive Cooperative Multiple Access in a Network with Two Classes of Users
In a network, most of the sources are bursty in nature, which leads to periods of silence in which users may have no data to transmit. These periods of silence result in under-utilization of the channel resources. Cooperative communications and spectrum sharing, originally considered as separate problems, are in fact dual problems in the sense that the available unused or under-used channel resources can be utilized to enhance the primary system performance via cooperation, or it can be shared by a secondary system to transmit new information. In this work, we try to investigate the possibility of dividing network free resources between secondary users to transmit their own information and relays to help both primary and secondary users.
Date: 7/11/2007
Speaker: Matt Stamm
Title:
Date: 6/27/2007
Speaker: Karim Seddik
Title: Trans-Modulation in Wireless Relay Networks
We consider the trans-modulation design for the decode-and-forward relay networks. We propose to reassign the constellation points at the relay nodes in such a way that the symbol error rate (SER) at the destination node is minimized. The proposed trans-modulation scheme can significantly improve the system SER performance without increasing the complexity of the system, especially when the relays are close to the source. For this case, improvements of about 2 dB for 16-QAM constellation and about 3 dB for 64-QAM constellation are achieved for the single relay case.
Title: Distributed Detection in Sensor Networks: a Sensor or a Relay
Several works have considered the design of distributed detection schemes in sensor networks. Most of these works have focused on having a network power or rate constraint. Most of them have focused on characterizing the performance in terms of "error exponent" (which is the rate of decay of some probability of error as the number of sensors or observations tends to infinity). In this talk, I will present some preliminary results on a different issue, whether it is better to use the sensor to sense or as a relay at a certain location in the network. The idea is that sensors that sense "less informative" data can better work as relays. In this talk, I will present the case of Gaussian observations over AWGN channels.
Date: 6/20/2007
Speaker: Peng Qiu
Title: P-spectrum for QRS Detection in EKG Signals
EKG stands for electrocardiogram. It is the signal of electrical voltage in the heart. One important piece of information we can get from the EKG signal is the Heart Rate Variability (HRV). HRV describes how the heart rate changes along time, or how the time interval between consecutive heart beats changes along time. Heart rate is controlled by the human automatic nervous system, which responds to stress, pain, and various other physical conditions. The ultimate goal of this study is to find correlation between a patient’s HRV and how much pain the patient is suffering, and to apply this correlation to assist diagnosis of children with asthma. As a first step toward our goal, we develop a method to detect heart beats from EKG signals, which is also referred as QRS detection in the literature. This is an old problem. In this talk, I’ll present a new solution based on p-spectrum. The results show that the proposed method yields excellent detection performance (99%). Although the performance is comparable with that of existing methods, the proposed method has its advantages, such as do not require predetermined parameters, do not make assumptions on shape of the EKG signal, do not require any training, and can operate in real-time.

### Spring 2007

Date: 5/9/2007
Speaker: Yao Ma (Iowa State University)
Title: Effects of Imperfect and Outdated Feedback on the Transmit Beamforming Performance and Countermeasures
Date: 4/25/2007
Speaker: Amr El Sherif
Title: Opportunistic Spectrum Sharing in Relay Networks
Cooperative communications and cognitive spectrum sharing are new techniques that emerged recently with the aim of efficient utilization of network resources. Cooperative communications allow different users to share their resources for distributed transmission of informations. On the other hand, cognitive spectrum sharing techniques allow a secondary system to opportunistically share the spectrum of the primary licensed users when it is idle. So far, these two pradigms have been studied independently. In this work, we try to investigate the possibility of dividing network free resources between relays for cooperative transmission and secondary users to transmit their own information. We characterize the stable throughput region of the new system and study the tradeoff between allocating resources for cooperation and spectrum sharing.
Date: 4/11/2007
Speaker: Steve Tjoa, Ahmed Ibrahim, Wan-Yi Lin
This afternoon, we will give our practice talks in preparation for ICASSP 2007.
Date: 4/4/2007
Title: Cross-Layer Design for Cooperative Communications and Networking
Cooperative communications is a new communication paradigm in which different terminals in the wireless network share their antennas and resources for distributed transmission and processing. Recent studies have shown that cooperative communications can yield significant performance improvement due to spatial diversity gains. The theory of cooperative communications is however still immature to fully understand its broader impacts on the design of future wireless networks. This thesis contributes to the advancement of cooperative communications by developing and analyzing cooperation protocols at different network levels, with the goal to provide significant improvements in signal reliability, coverage area, network throughput, and energy efficiency with respect to other existing alternatives.
Date: 3/28/2007
Speaker: Visa Koivunen (Helsinki University of Technology, Finland)
Date: 3/14/2007
Speaker: Ahmed Ibrahim
Title: Network Maintenance in Wireless Sensor Networks
Network connectivity is necessary for successful communication among the network's nodes. In this paper, we address the problem of network maintenance in wireless sensor networks, i.e., improving the network connectivity by adding an arbitrary set of relay nodes to it. In particular, we try to find the optimum locations for these relays in order to maximize the network connectivity. We characterize the network connectivity by the Fiedler value, which is the second smallest eigenvalue of the Laplacian matrix representing the network graph. We formulate a relaxed version of of this problem as a semi-definite programming problem (SDP) and find a near-optimum solution to it. We show that the proposed algorithm enhances the Fiedler value by 60% due to adding one relay only. Finally, we show that our proposed algorithm can also be used in some cases to find the optimum location for a single relay in order to reconnect a disconnect network.
Date: 3/7/2007
Speaker: Steve Tjoa
Title: Cut-and-Paste Forensics: An Optimization-Based Framework to Detect Digital Image Tampering
Image segmentation is an area of research that has existed for a few decades now, with applications covering many aspects of computer vision and pattern recognition. Today, we discuss another problem which is intimately related to image segmentation, namely, detection of cut-and-paste forgeries in digital images. Recent work in the computer vision field has resulted in new optimization-based approaches for achieving image segmentation which depart from traditional approaches. Drawing from this body of work along with our own recent work in source coding forensics, we offer an optimization-based method for detecting cut-and-paste forgeries based upon compression-related feature selection. We will present a thorough discussion of the problem formulation along with results.
Date: 2/28/2007
Speaker: Karim Seddik
Title: Distributed Space-Frequency Coding over Relay Channels
Designing diversity achieving schemes over the wireless multi-path fading relay channels is crucial to achieve high performance gains. These gains are achieved by exploiting the multi-path (frequency) and cooperative diversities to combat the fading nature of wireless channels. In this talk, the design of distributed space-frequency codes (DSFCs) for wireless relay networks is considered. The term distributed comes from the fact that the space-frequency code is distributed among randomly located relay nodes. The proposed DSFCs are designed to achieve maximum diversity over the wireless relay channels. A two-hop system model, where there is no direct link from the source node to the destination node, is considered. We consider the use of DSFCs with the decode-and-forward (DAF) and amplify-and-forward (AAF) protocols. The code design criteria to achieve full diversity and maximum coding gain based on minimizing the pairwise error probability (PEP) are derived. For DSFC with the DAF protocol, the proposed DSFCs are proved to achieve full diversity of order LN where N is the number of relay nodes and L is the number of paths per channel. For DSFC with the AAF protocol, the proposed DSFCs are proved to achieve full diversity of order LN for any number of relays, N, for the special cases of L=1 (flat fading) and L=2 (two-rays channel model).
Date: 2/21/2007
Speaker: Beibei Wang
Title: Primary-Prioritized Markov Approach for Dynamic Spectrum Allocation
Dynamic spectrum access has become a promising approach to fully utilize the scarce spectrum resources. In a dynamically changing spectrum environment, it is very important to consider the statistics of different users' spectrum access so as to achieve more efficient spectrum allocation. In this paper, we propose a primary-prioritized Markov approach for dynamic spectrum access through modeling the interactions between the primary and the secondary users as continuous-time Markov chains (CTMC). Based on the CTMC models, to compensate the throughput degradation due to the interference among unlicensed users, we derive the optimal access probabilities for the unlicensed users, by which the spectrum access of the unlicensed users is optimally coordinated, and the spectrum dynamics are clearly captured. Therefore, a good tradeoff can be achieved between the spectrum efficiency and fairness. The simulation results show that the proposed primary-prioritized dynamic spectrum access approach under proportional fairness criterion achieves much higher throughput than the CSMA-based random access approaches and the approach achieving max-min fairness. Moreover, it provides fair spectrum sharing among unlicensed users with only small performance degradation compared to the approach maximizing the overall average throughput.
Date: 2/7/2007
Speaker: Peng Qiu
Title: Dependence Model and Network for Cancer Classification, Prediction and Biomarker Identification
In recent years, high throughput measuring techniques (gene microarrays, protein mass spectrum) have made it possible to simultaneously monitor the expression of thousands of genes or proteins. A topic of great interest is to study the difference of gene/protein expressions between normal and cancer subjects. In literature, various clustering methods have been proposed to analyze gene/protein data, and they are dominantly data-driven. In our work, we proposed an alternative model-driven approach. Our aim is to develop statistical models to systematically interpret the high throughput experiment data and reveal biology insights. We propose a dependence model which can be used to examine the interactions among genes/proteins: we can zoom out to study the big picture, the ensemble dependence relationships among groups of genes/proteins; we can also zoom in to examine the details, the relationship among individual genes/proteins. We have shown that the dependence model is highly effective in the classification of data from normal and cancer subjects. We discovered that the eigenvalue of the dependence model exhibits different patterns for normal and subjects at different stages of cancer development. This indicates the dependence model has the potential to predict cancer development. The concept of dependence network is proposed based on the dependence model. The interaction relationships among genes/proteins are modeled by the dependence network, from which we are able to reliably identify biomarkers, important genes/proteins for the early prediction and effective treatment of cancer.
Date: 1/31/2007
Speaker: Zhu Ji
Title: Optimal Pricing-Based Dynamic Spectrum Sharing
Dynamic spectrum allocation becomes a promising approach to increase the spectrum efficiency for wireless networks. However, the collusion among selfish network users may seriously deteriorate the efficiency of dynamic spectrum sharing. In this paper, we model the spectrum allocation in wireless networks with multiple selfish legacy spectrum holders and unlicensed users as multi-stage dynamic games. In order to combat user collusion, we propose a pricing-based collusion-resistant dynamic spectrum allocation approach to optimize overall spectrum efficiency while not only keeping the participating incentives of the selfish users but also combating possible user collusion. The simulation results show that our proposed scheme achieves high efficiency of spectrum usage even with the presence of severe user collusion.

### Fall 2006

Date: 12/13/2006
Speaker: Peng Qiu
Title: Dependence Model and Network for Classification and Biomarker Identification
Our purpose is to develop a statistical modeling approach for cancer biomarker discovery and provide new insights into early cancer detection. We propose the concept of dependence network, apply it for identifying cancer biomarkers, and study the difference between the protein or gene samples from cancer and non-cancer subjects based on mass-spectrometry (MS) and microarray data. Three MS and two gene microarray datasets are studied. Clear differences are observed in the dependence networks for cancer and non-cancer samples. Protein/gene features are examined three at one time through an exhaustive search. Dependence networks are constructed by binding triples identified by the eigenvalue pattern of the dependence model, and are further compared to identify cancer biomarkers. Such dependence-network-based biomarkers show much greater consistency under 10-fold cross-validation than the classification-performance-based biomarkers. Furthermore, the biological relevance of the dependence-network-based biomarkers using microarray data is discussed. The proposed scheme is shown promising for cancer diagnosis and prediction.
Date: 11/15/2006
Speaker: Steve Tjoa
Title: Digital Image Source Coder Identification: Transform Coder Classification
Traditional approaches to multimedia security (e.g. watermarking, cryptography) protect content using additive operations. With these methods, we require access to the original host signal. However, in many scenarios, we may not even have access to the host signal, and therefore we cannot enforce protection through any extrinsic means. With non-intrusive forensic analysis, the forensic analyst only has access to an output signal in a raw format, without any header information or metadata.
In this talk, we propose a method to identify the presence of source coding performed upon a digital image. By analyzing the artifacts left behind, we can identify the history of operations, including pre-processing, transformation, and quantization. We employ a probability-based distance measure to classify the method of transform coding previously performed upon the image, and we show results across a variety of transform methods, quality factors, and additive noise.
Date: 11/8/2006
Speaker: Amr El Sherif
Title: Cooperation During Silence Periods in Packet Voice Communications
Speech communication is characterized by periods of silence in between talk spurts that account for roughly 60 % of the conversation time. This is a key property that, when exploited, could significantly improve the utilization of channel resources but at the cost of requiring a more sophisticated multiple access protocol. For wireless channels, these issues are addressed in the Packet Reservation Multiple Access (PRMA) protocol. Unfortunately, transmission errors inherent to wireless communications significantly impact the performance of PRMA. To mitigate the negative channel effects on the protocol, we propose the deployment of a relay terminal in the PRMA network. The relay will utilize a fraction of the free resources available in the network (due to silence periods) to introduce spatial diversity without sacrificing the bandwidth efficiency of the system. Analytical and numerical results reveal a significant improvement in the network throughput and the perceived speech quailty over the non-cooperative protocol.
Date: 10/25/2006
Speaker: Wan-Yi Lin
Title: Multi-User Collusion Behavior Forensics: Game Theoretic Formulation of Fairness Dynamics
Multi-user collusion is a cost-effective attack against digital fingerprinting, in which a group of attackers collectively undermine the traitor tracing capability of digital fingerprints. However, during multi-user collusion, each colluder wishes to minimize his/her own risk and maximize his/her own profit, and different colluders have different objectives. Thus, an important issue during collusion is to agree on how to distribute the risk/profit among colluders and ensure fairness of the attack. To have a better understanding of the attackers' behavior during collusion to achieve fairness, we model the dynamics among colluders as a non-cooperative game. We then study the Pareto-Optimal set, where no colluder can further increase his/her own payoff without decreasing others', and analyze the Nash Bargaining solution, and also other solutions of this game.
Date: 10/18/2006
Speaker: Beibei Wang
Title: Opportunistic Spectrum Access for Unlicensed Users With Prioritized Primary Access
Today's wireless networks are regulated by a fixed spectrum assignment policy, which doesn't efficiently utilize the limited available spectrum resources. Dynamic spectrum access is proposed to solve these current spectrum inefficiency problems. Among branches of the dynamic spectrum access, secondary usage of the licensed spectrum is a viable solution to more efficient utilization of the frequency spectrum. This means, when certain spectrum is not currently used by any primary user, unlicensed users share this spectrum opportunistically. However, when primary users re-appear, the unlicensed users on the same spectrum should vacate. How to combine the primary user activity for opportunistic spectrum allocation is still an open issue. In this work, a continuous time Markov chain (CTMC) is presented to model the dynamics of the system's evolvements. The limiting behaviors of the primary user and unlicensed users are characterized. A proportional fair (PF) opportunistic spectrum access scheme is proposed to efficiently and fairly assign the spectrum access to the unlicensed users. Finally, simulation studies are provided to justify the proposed scheme.
Date: 10/4/2006
Speaker: Ahmed Ibrahim
Title: Distributed Power-Efficient Cooperative Routing in Wireless Ad Hoc Networks
Recently, the merits of the cooperative communications in the physical layer have been explored. However, the impact of the cooperative communications on the design of the higher layers is not well-understood yet. In this paper, we consider the power-efficient cooperative routing problem in wireless ad hoc networks, which exploits the features of the wireless medium. We propose the Minimum Power Cooperative Routing (MPCR) algorithm, which requires polynomial complexity. The MPCR algorithm constructs the minimum-power route as a cascade of the minimum-power single-relay building blocks from the source to the destination. Thus, any distributed routing algorithm can be utilized to find the optimal route with minimum end-to-end transmitted power, while guaranteeing certain throughput. The simulations show that the MPCR algorithm can achieve power saving of 9.3% compared to the existing cooperative routing algorithm, where the selected routes are constructed based on the noncooperative routes.
Date: 9/20/2006
Speaker: Karim Seddik
Title: Source-Channel Diversity for Multi-Hop and Relay Channels
A key challenge in the design of real-time wireless multimedia systems is the presence of fading coupled with strict delay constraints. A very effective answer to this problem is the use of diversity achieving techniques. This talk focuses on studying systems that exhibit diversity of three forms: source coding diversity, channel coding diversity and user-cooperation diversity (implemented through either relay channels or multi-hop channels, each with amplify-and-forward or decode-and-forward user cooperation). Consistent with the focus on real-time multimedia communications, performance is measured through the distortion exponent, which measures the rate of decay of the end-to-end distortion at high SNRs. The results show that for both relay and multi-hop channels, optimum channel coding diversity provides the best performance, followed by source coding diversity. The results also show a tradeoff between the quality (resolution) of the source encoder and the amount of cooperation, in that, as the bandwidth expansion factor increases (higher bandwidth) user cooperation diversity is the main limiting factor, not the source encoding distortion. Thus, the distortion exponent is improved by increasing the number of relays (increasing the diversity order). At low bandwidth expansion factor the source average end-to-end distortion is limited by the source encoder distortion and, in this case, using higher resolution source encoder will improve the performance (in terms of the distortion exponent) more than increasing the number of relay nodes.
Date: 9/13/2006
Speaker: Charles Pandana
Title: Spectral Graph Information in Sensor Networks
There are many important criteria in designing algorithms in sensor network, namely energy efficiency, energy awareness, topology awareness, and connectivity awareness, etc. Among all those criteria, the network connectivity is a very important one, since most of the sensors in the sensor network are static and they stay in their original position for the entire of their lifetime. Therefore, it is important to design a robust metric that characterizes the network connectivity. In this talk, we will explain how to model the connectivity of the network using the spectral of the graph. We propose link models. We also provide 2 applications, namely routing and network health maintenance. In both applications, we show how the spectral of graph can be used to derive the connectivity-aware solution.
Date: 9/6/2006
Title: Possible Gains of Cooperation When Applied to Sensor Networks
The gains of cooperative communications in wireless networks have been explored recently under the ideal assumption of negligible receiving and processing power. In sensor networks, the power spent for listening and computing can constitute a significant portion of the total consumed power, and such an overhead can reduce the gains promised by cooperation. In this work, cooperation gains are investigated by taking into consideration such overheads in the analytical framework. The performance metric considered is the energy efficiency of the system measured by the total power required to achieve a certain quality of service requirement. The analytical and numerical results reveal very interesting threshold behavior below which direct transmission is more energy efficient, and above which cooperation provides more gains. Such a tradeoff is shown to depend on many parameters such as the relative locations of the source and destination, the values of the receive and processing powers, the application, and many other factors. Moreover, there are experimental results conducted to verify the channel model.

### Summer 2006

Date: 8/22/2006
Speaker: Amr El Sherif
Title: Cooperative Packet Reservation Multiple Access Protocol
PRMA is an efficient statistical multiplexing scheme which combines random access with time division. It allows a group of spatially dispersed terminals to transmit packetized voice to a common base station over a shared channel. In this scheme, transmission errors have a significant impact on the system capacity and speech quality. In this work, we add user cooperation capability to the PRMA protocol to improve system performance. We exploit the on/off characteristic of speech to allow some of the users to act as relays for other users. In the talk, a detailed Markov analysis of our proposed protocol will be presented, using this analysis we are able to compute system capacity, access delay and packet dropping probability.
Date: 8/15/2006
Speaker: Steve Tjoa
Title: Digital Image Source Coder Identification: A Non-Intrusive Forensic Methodology
Multimedia security, in all its forms, is a broad field devoted to protecting the content of data through such approaches as cryptography, watermarking, fingerprinting, and network security. However, in the absence of such security schemes, there does not exist any artificial signals or operations added to the image to enforce protection. Today, we explore a growing field of multimedia forensics known as non-intrusive forensic analysis, where the forensic analyst only has access to the output data in a raw format, without any header information or metadata. In performing forensic analysis, we can achieve many objectives such as datapath integrity and patent enforcement. We present a non-intrusive forensic methodology related to the identification of source coding for images, including the detection of pre-processing, transform method, and quantization method. This methodology can be applied to a wide variety of images encoded using one of many families of encoders, including transform-based, predictive, vector quantization, and fractal coders.
Date: 8/8/2006
Speaker: Beibei Wang
Title: Distributed Relay Selection and Power Control for Multiuser Cooperative Communication Networks Using Buyer/Seller Game
The performances in cooperative communications depend on careful resource allocation such as relay selection and power control, but traditional centralized resource allocation needs considerable overhead and signaling to exchange the information for channel estimations. In this paper, we propose a distributed buyer/seller game theoretic framework over multiuser cooperative communication networks to stimulate cooperation and improve the system performance. By employing a two-level game to jointly consider the benefits of source nodes as buyers and relay nodes as sellers, the proposed approach not only helps the source smartly find the relays at relatively better locations and buy optimal amount of power from them, but also helps the competing relays maximize their own utilities by asking the reasonable prices. The game is proved to converge to a unique optimal equilibrium. From the simulation results, the relays in good locations can play more important roles in increasing source node's utility, so the source would like to buy more power from these preferred relays. On the other hand, the relays have to set the proper prices to attract the source's buying because of competition from other relays and selections from the source. Moreover, the distributed game theoretic resource allocation can achieve comparable performance compared with the centralized one.
Date: 8/1/2006
Speaker: Wan-Yi Lin
Title: Image Coding Forensics
In many signal processing, it's important to determine if the source is the desired one. In this paper, we'll present a method to identify the source coding scheme given a distorted picture. Our system can achieve identification rate over 95% for a PSNR between 20 dB and 35 dB, and identification rate over 80% for a PSNR between 35dB and 40dB. Also our system gives a confidence measure of our result.
Date: 7/25/2006
Speaker: Zhu Ji
Title: Pricing-Based Dynamic Spectrum Allocation against User Collusion in Wireless Networks with Selfish Users
In order to fully utilize the scarce spectrum resources, with the development of cognitive radio technologies, dynamic spectrum allocation becomes a promising approach to increase the efficiency of spectrum usage. In this paper, we consider the spectrum allocation in wireless networks with multiple selfish legacy spectrum holders and unlicensed users as multi-stage dynamic games. By studying the selfish users' collusive cheating behaviors, a robust and efficient dynamic pricing approach is proposed to optimize overall spectrum efficiency while not only keeping the participating incentives of the selfish users but also combating possible user collusion. The simulation results show that our proposed scheme achieves high efficiency of spectrum usage even with the presence of severe user collusion.
Date: 7/18/2006
Speaker: Ahmed Ibrahim
Title: QoS Minimum-Power Cooperative Routing for Cluster-based Sensor Networks
We study the impact of the new cooperative communications paradigm on the routing problem in wireless sensor networks. We consider the minimum-power routing problem with desired Quality of Service (QoS). The QoS, characterized by the end-to-end probability of success and bandwidth efficiency, represents the reliability and capability of the route. We propose a routing metric, which considers the merits of the cooperative communications to obtain the required transmitted power in order to achieve certain QoS. Since the minimum-power routing problem is an NP-complete problem, we propose simple heuristic routing algorithms, namely, Minimum Power Cooperative Routing (MPCR) and Cooperation Along the Shortest Non-Cooperative Path (CASNCP) algorithms, which require less power compared to the conventional shortest-path routing algorithms. The numerical results indicate that the proposed algorithms reduce the required power by up to 21% compared to the conventional shortest-path routing algorithms.

### Spring 2006

Date: 5/1/2006
Speaker: Steve Tjoa
Title: Non-Intrusive Forensics for Image Source Coding
We present forensic issues related to image source coding, including the detection of blocking artifacts, detection of transform method, quantization step estimation, and PSNR estimation. In performing forensic analysis, we can achieve many objectives such as quality assurance of transmitted multimedia, rate control, signal restoration, security and authentication, and patent enforcement. This talk focuses on parameter estimation, using JPEG and SPIHT as example source coders. For JPEG, our system almost always accurately estimates the quantization table without error, and estimates the PSNR to within an error of 0.92 dB. An analytical formulation for SPIHT PSNR estimation employing the Cauchy distribution is also discussed. Combining results, we create a system which identifies what source coder was used to encode an image, along with the coding parameters.
Date: 4/24/2006
Speaker: Andres Kwasinski
Title: Performance Modeling and Analysis of Multimedia Systems with Joint Source-Channel Coding and its Application to Source-Channel-Cooperation Tradeoffs
The study of Joint Source-Channel Coding bit rate allocation faces the challenge of obtaining a close form solution for the function that links end-to-end distortion with channel signal-to-noise ratio when considering non-capacity achieving source and channel codes. In this talk, we start by addressing this problem by characterizing the Distortion-SNR (D-SNR) curve through a representation that uses a set of carefully selected points. I will show that the D-SNR can be closely approximated through a linear function in log-log scales and we discuss the implications of this result. Next, we use this approach to study the tradeoffs between Joint Source-Channel Coding bit rate allocation and the use of user cooperation. In the talk, I will focus on studying the effects that the source-channel-cooperation tradeoffs have on the D-SNR performance of different schemes and how this translates into conditions where use of cooperation present an advantage or not.
Date: 4/17/2006
Speaker: Zhu Ji
Title: Belief-Assisted Pricing for Dynamic Spectrum Allocation in Wireless Networks with Selfish Users
In order to fully utilize the scarce spectrum resources, with the development of cognitive radio technologies, dynamic spectrum allocation becomes a promising approach to increase the efficiency of spectrum usage. In this paper, we consider the spectrum allocation in wireless networks with multiple selfish legacy spectrum holders and unlicensed users as multi-stage dynamic games. A belief-assisted dynamic pricing approach is proposed to optimize overall spectrum efficiency while keeping the participating incentives of the users based on double-auction rules. Moreover, considering the budget constraints of the unlicensed users, a dynamic programming approach is further developed to optimize the spectrum allocation over time. The simulation results show that our proposed scheme not only approaches optimal outcomes with low overhead compared to general continuous double auction mechanisms, but also fully exploits the time diversity of spectrum resources when budget constraints exist.
Date: 4/10/2006
Speaker: Quoc Lai
Title: Baseband Implementation and Performance Analysis of the Multiband OFDM UWB System
In April 2002, the U.S. Federal Communications Commission (FCC) published a report and order that allows Ultra-WideBand (UWB) communication systems to be deployed on an unlicensed basis following Part 15 rules. The publication encourages researchers and scientists to devote their time and resources to the development of the UWB transmission technology. Due to its large transmission bandwidth, the UWB transmission technology promises to fulfill the demand of high transmission data rates and is considered as one of the transmission technologies in the coming fourth generation (4G) of wireless transmission. In this thesis, we follow the standard proposal IEEE 802.15.3a to implement the multiband OFDM UWB system in C programming language. After that, we analyze the system performance in the multipath channel, the IEEE 802.15.3a channel standard. Four different scenarios relating to the imperfection of frequency and timing synchronization are considered. The analysis is performed in two stages. First, we consider the performance of only the OFDM system. Degradation ratio and average bit error probability are the two metrics used to evaluate the performance. Secondly, we provide the way to obtain the performance bound of the entire MB-OFDM UWB system. The performance analysis provides us a good understanding of the system behavior in the IEEE 802.15.3a channel standard under various synchronization conditions. The knowledge obtained from the performance analysis would help improving the MB-OFDM UWB system in future, and that is our main contribution to the area of wireless communications.
Date: 3/27/2006
Speaker: Thanongsak Himsoon
Title: Differential Modulation with Threshold-Based Decision Combining for Decode-and-Forward Cooperative Communications
Differential modulation is widely known as a practical alternative that provides a good tradeoff between receiver complexity and performance. However, the available differential schemes for wireless relay networks require perfect synchronization and/or provide limited transmission rates. This paper proposes a threshold-based differential decode-and-forward cooperative scheme that not only alleviates the problems of synchronization and rate limitation, but also efficiently exploits the cooperative relay channels via the use of a pre-determined decision threshold. In the proposed scheme, the source information is forwarded by the relay only if it is correctly decoded. The properly-designed threshold enables the destination to decide whether the received signal from the relay contains information such that the received signals from the source and the relay can be efficiently combined and jointly decoded. The bit error rate (BER) performance analysis of the proposed scheme is analyzed in case of differential M-ary PSK signals. A tight BER approximation is established, and BER upper bound and lower bound are determined. Based on the tight BER approximation, joint optimum decision threshold and power allocation are numerically evaluated. Both analytical and simulation results reveal that the decision threshold and the power allocation depend on qualities of the communication channels. Interestingly, when the link quality between relay and destination is very good, the effect of the threshold dominates the effect of the power allocation at high signal-to-noise ratio. For example, in case of differential QPSK signals with equal power allocation, the proposed scheme with a properly-designed threshold leads to more than 5 dB performance improvement over the scheme without the threshold at a BER of 104. When the transmitted power is allocated optimally, the performance is further improved by 0.5 dB at the same BER. Extensive simulation results are provided to validate the merit of the proposed scheme and confirm the theoretical analysis.
Date: 3/13/2006
Speaker: Peng Qiu
Title: 3D contour reconstruction for radiation therapy treatment
Radiation therapy treatment (i.e. Gamma Knife) is a kind of non-invasive, painless, bloodless surgery. During the treatment, hunderds of low energy radiation beams are focused on the tumors (or unwanted tissues), to kill them. Before the surgery, one important thing is to know exactly the position, size and shape of the tumor, through some medical imaging techniques. This is easy when the tumor is standing still, not moving at all. However, when the tumor is located in the chest, because of the breathing, the tumor is moving all the time. Ignoring the movement will cause lots of artifacts in the estimation of the tumor's geometry. In this talk, I will present the full story and present a method about how to deal with the movement.
Date: 3/6/2006
Speaker: Beibei Wang
Title: Stackelberg Game for Distributed Resource Allocation over Multiuser Cooperative Communication Networks
In this talk, we propose a Stackelberg game theoretic framework for distributive resource allocation over multiuser cooperative communication networks to improve the system performance and stimulate cooperation. Two questions of who should relay and how much power for relaying are answered, by employing a two-level game to jointly consider the benefits of source nodes as buyers and relay nodes as sellers in cooperative communication. From the derived results, the proposed game not only helps the source smartly find relays at relatively better locations but also helps the competing relays ask reasonable prices to maximize their own utilities. From the simulation results, the relays in good locations or good channel conditions can play more important roles in increasing source node's utility, so the source would like to buy power from these preferred relays. On the other hand, because of competition from other relays and selections from the source, the relays have to set the proper prices to attract the source's buying so as to optimize their utility values.
Date: 2/27/2006
Speaker: Wei Yu
Title: Cooperation Stimulation and Security in Autonomous Mobile Ad Hoc Networks
In autonomous mobile ad hoc networks, nodes belong to different authorities and pursue different goals, therefore cooperation among them cannot be taken for granted. Meanwhile, some nodes may be malicious whose objective is to damage the network. In this talk I will present a joint analysis of cooperation stimulation and security in autonomous mobile ad hoc networks under a game theoretic framework. After that, I will present the progress of our sensor network and collaborative communication testbed implementation.
Date: 2/20/2006
Title: Cognitive-Based Cooperative Multiple-Access Protocols
In today's groupmeeting, I will continue talking about the "Cognitive based Cooperative Multiple-Access" protocol I proposed last summer. The work is almost complete now and I have proven rigorously the results I conjectured before regarding the stability region. Moreover, I included a delay-analysis for the two-user symmetric case which is, beside queue stability, another very important performance measure for a network. I also included a comparison to conventional cooperative transmission and some ideas for future work.
Date: 2/13/2006
Speaker: Wan-Yi Lin
Title: Colluder-Detector Behavior Forensics in Scalable Multimedia
Digital fingerprinting uniquely labels each distributed copy with user's ID and provides a proactive means to track the distributionof multimedia. Multi-user collusion is a powerful attack against digital fingerprinting, in which a group of attackers collectively mount attacks to remove the embedded identification information. To resist such multi-user collusion and support multimedia forensics, we investigate the side information based multimedia fingerprinting. We explore techniques to utilize side information of collusion attacks during colluder identification process, and show that the means of the detection statistics at the detector's side can significantly improve the traitor tracing capability. We also investigate how the fingerprint detector can probe such side information from the colluded copy, and our simulation results show that the proposed scheme helps the fingerprint detector achieve the optimum detection performance.
Date: 2/6/2006
Speaker: Karim Seddik
Title: Protocol-Aware Design Criteria and Performance Analysis for Distributed Space-Time Coding
In this talk, we consider the design of distributed space-time codes for wireless networks. Distributed space-time coding (DSTC) can be achieved through node cooperation to emulate multiple transmit antennas. We derive the distributed space-time codes design criteria for different scenarios based on the pairwise error probability (PEP) analysis. First, we consider the decode-and-forward protocol in which each relay node decodes the symbols that it receives from the source node before retransmission. We prove that space-time codes, designed to achieve full diversity and maximum coding gain in the MIMO channels, will achieve full diversity but not necessarily maximizing the coding gain if used with the decode-and-forward protocol. Next we consider the amplify-and-forward protocol in which each relay node does not decode the symbols but it can perform simple operations such as linear transformation of the received signals and then amplifies the signal before retransmission. We derive the space-time code design criteria for the amplify-and-forward protocol and prove that a space-time code designed to achieve full diversity and maximum coding gain in MIMO channels will achieve the same if used with the amplify-and-forward protocol.
Date: 1/30/2006
Speaker: Hong Zhao
Title: Multimedia Forensics for Secure Multi-user Communications
Recent development in multimedia and network technologies has raised the critical issue of protecting multimedia content security and enforcing digital rights. Cryptographic tools (e.g., encryption and authentication) alone are not sufficient since the protection disappears after the data are delivered to users. To address the post-delivery protection, multimedia forensics combines digital domain evidence to determine whether multimedia content has been altered, to indicate how these alterations were made, and to identify who participated in the alterations.
This talk addresses several issues in multimedia forensics. First, multimedia fingerprinting for traitor tracing will be presented. Then, the dynamics among users in the forensic systems and the behavior forensics will be formulated and analyzed. Finally, the impact of forensic tools on current multimedia communication frameworks will be investigated, and secure fingerprint multicast schemes will be presented, which collectively address the security and the bandwidth efficiency of networked multimedia.
Date: 1/23/2006
Speaker: Zhu Han
Title: How to Cooperate in Wireless Networks?
Wireless networking and resource allocation are general strategies to utilize the limited wireless radio resources, control the co-channel interferences, and enhance the network performances. To reduce the overhead imposed by such strategies, the mobiles of the next generation networks have their own autonomies for resource allocation in a distributive way. However, the non-cooperative competition of the radio resources results in low system efficiency. So how to ensure cooperation among autonomous users is one of the most important wireless networking research topics. There are many types of approaches to enforcing cooperation such as incentive based, referee based, punishment based, collaborative communication based, etc. In this talk, we concentrate on one of the approaches and employ it to OFDMA networks.

For incentive based approach, we propose bargaining method to have mutual benefits. Specifically, a fair and simple scheme to allocate subcarrier, rate, and power for multiuser single cell OFDMA systems is considered. The problem is to maximize the overall system rate, under each user's maximal power and minimal rate constraints, while considering the fairness among users. The approach proposes the fairness and low complexity implementation based on Nash Bargaining Solutions and Coalitions. First, a two-user algorithm is developed to bargain subcarrier usage between both users. Based on this algorithm, we develop a multiuser bargaining algorithm where optimal coalition pairs among users are constructed. The simulation results show that the proposed algorithms not only provide fair resource allocation among users, but also have comparable overall system rate with the scheme maximizing the total rate without considering fairness. They also have much higher rates than those of the scheme with max-min fairness.

We also briefly mention the other approaches. For referee based approach, we propose non-cooperative game approach with a referee for multi-cell OFDMA networks. Collaborative communication based approach and punishment based approach are also briefly discussed to give an overview on how to cooperate in wireless networks.

### Fall 2005

Date: 12/6/2005
Speaker: Peng Qiu
Title: Resynchronization Analysis of Cell-cycle Gene Expression Data
Identification of genes expressed in a cell-cycle-specific periodical manner is of great interest to understand cyclic systems which play a critical role in many biological processes. However, identification of cell-cycle regulated genes by raw microarray gene expression data directly is complicated by the factor of synchronization loss, thus remains a challenging problem. Decomposing the expression measurements and extracting synchronized expression will allow to better represent the single-cell behavior and improve the accuracy in identifying periodically expressed genes. In our study, we propose a resynchronization-based algorithm for identifying cell-cycle-related genes. We introduce a synchronization loss model by modeling the gene expression measurements as a superposition of different cell populations growing at different rates. The underlying expression profile is then reconstructed through resynchronization and is further fitted to the measurements in order to identify periodically expressed genes. Results from both simulations and real mircorarray data will be showed.
Title: mPCA Model for Clustering in PET Parametric Image
mPCA stands for mixture principle component analysis. It combines a probabilistic formulation to PCA, makes it a mixture model and brings in the Bayesian framework. mPCA can be used as a classification tool, both for supervised and unsupervised learning. In our study, we use mPCA as an unsupervised clustering tool and apply it in PET parametric imaging. Through mPCA, we segmented the PET image into different parts, each corresponds to tissues with different properties. I’ll explain what is the “different properties?during the talk.
Date: 11/22/2005
Speaker: Zhu Ji
Title: Cooperation Enforcement in Autonomous Ad Hoc Networks Under Noisy and Imperfect Observation
In autonomous mobile ad hoc networks (MANET) where each user is its own authority, fully cooperative behaviors, such as unconditionally forwarding packets for each other, cannot be directly assumed. In this talk, we focus on cooperation enforcement in autonomous mobile ad hoc networks under noisy and imperfect observations and study the basic packet-forwarding function using the repeated game models with imperfect information. Two approaches have been proposed to obtain cooperation-enforcement strategies based solely on each node's own past actions and its private imperfect observation of other nodes' information: belief-based packet forwarding framework and review-based packet forwarding framework. As for the belief-based approach, we not only show that the proposed strategy with belief system can achieve sequential equilibrium but also establish its performance bounds. The simulation results illustrate that the proposed belief-based packet forwarding approach can enforce the cooperation with only a small performance degradation compared to the unconditionally cooperative outcomes. As for the review-based strategy, we show that this approach is able to approach the unconditional cooperative outcomes by using large-enough review length. Simulation results match our theoretical derivation.
Date: 11/1/2005
Speaker: Karim Seddik
Title: Outage Analysis of Multi-node Amplify-and-Forward Relay Networks
We consider the outage probability analysis of multi-node amplify-and-forward relay network with N relay nodes helping the source. We consider a system in which each relay node amplifies the source signal only. We obtain an approximation for the outage probability which is tight at high signal-to-noise ratio (SNR). This tight outage approximation shows that the system can achieve a maximum diversity of order N+1. For the case of N=1, our approach gives the same result obtained previously by Laneman et. al. for the single relay scenario.
Title: Distributed Diagonal Space-Time Codes in Wireless Relay Networks
We consider the design of distributed diagonal space-time codes (DDSTC) for N relay nodes helping the source. We impose the diagonal structure of the code to simplify synchronization between the different relay nodes because it is very difficult to synchronize simultaneous transmissions of randomly distributed relay nodes. We derive an upper bound on the outage probability of the system, which shows that a diversity of order N can be achieved. Then, we derive the code design criterion for the DDSTC based on minimizing the pairwise error probability (PEP). It turns out that the code design criterion is to maximize the minimum product distance, which is the same criterion used for designing diagonal algebraic space-time (DAST) codes and designing full-rate full-diversity space frequency (SF) codes.
Date: 10/25/2005
Speaker: Charles Pandana
Title: Robust Maximum Connectivity Energy-Aware Routing for Wireless Networks
In typical sensor network deployment, some nodes may be more important than other nodes because the death of these nodes cause the network disintegration. The network disintegration causes early termination of information delivery. To overcome this problem, we propose a class of routing algorithms called keep connect algorithms, that explicitly consider the connectivity of the network while making the routing decision. The algorithm can be used along with any existing routing algorithms. When doing the routing decision, the keep connect algorithm embeds the importance of the nodes in the routing cost. The importance of a node is quantified by the connectivity of the remaining network when that particular node dies. In particular, we propose two criteria for describing the connectivity of the remaining network. First, the importance of a node is quantified by how severe the remaining network becomes disconnected/disintegrated when that particular node dies. Second, the connectivity of the remaining network is quantified by the Fiedler value of the graph when that particular node is removed. The proposed algorithm achieves on average 20 ~ 50% better network lifetime and total delivered packets when it is used on top of MTE algorithm. The proposed algorithm also achieves around 20% improvement compared to the flow augmentation algorithm. We also outline the distributed implementation of our proposed algorithm. The MTE based distributed implementation achieves more than 2 times more total delivered packets before the network becomes disconnected, compared to flow augmentation based algorithm.
Title: Distributed Cooperative Routing Algorithms for Maximizing Network Lifetime
Date: 10/18/2005
Title: New Collaborative Multiple Access Protocols for Wireless Networks
In this work, a new multiple access approach is proposed that takes into account the broadcast nature of the wireless channel. The new approach employs a relay to boost the system throughput. This approach is based on a new idea in which the relay utilizes the empty time slots available in a TDMA frame. The relay stores the packets that failed transmissions in previous time slots. At each time slot, the relay listens to the channel and retransmits the packet at the head of its queue if the channel is free. This will better utilize the channel resources and will introduce on-demand spatial diversity into the network. Two different protocols are proposed to implement this new multiple-access scheme. The stability criteria of the associated queueing systems are studied and analytical expressions for the maximum stable throughput are provided for the symmetrical users case. Numerical results indicate a significant increase in the maximum stable throughput by using the new multiple-access protocol over pure TDMA.
Date: 10/11/2005
Speaker: Ahmed Ibrahim
Title: Multi-Node Cooperative Communications with Relay-Selection: When to Cooperate and Whom to Cooperate with?
In this paper, we propose a new cooperative communication protocol which achieves higher data rate while keeping the same diversity order as that of the conventional cooperative schemes. The proposed scheme considers relay selection via the available partial channel state information (CSI) at the source and the relays. In this work, we discuss the single and multi relay-selection decode-and-forward cooperative scenarios.

In the single-relay scenario, we determine when the source needs to cooperate with the relay, i.e., When to cooperate?''. We prove that full diversity order is achieved and show that the data rate is significantly increased. Furthermore, we derive an approximate expression of the achievable data rate and an upper bound on the symbol error rate (SER) performance for M-PSK signalling.

In the multi-node scenario, where arbitrary N relays are available, the source determines when it needs to cooperate with one relay only, and which relay to cooperate with, i.e., When to cooperate?'' and Whom to cooperate with?''. We show that full diversity is guaranteed and that a significant increase of the data rate is achieved. For the symmetric scenario, we derive an approximated data rate expression and obtain an upper bound on the SER performance. Moreover, we present the tradeoff between the achievable data rate and the corresponding SER. Finally, the obtained analytical results are verified through computer simulations.

Date: 10/4/2005
Speaker: Andres Kwasinski
Title: Cooperative Multimedia Communications: Joint Source Coding and Collaboration
Cooperative diversity exploits the broadcast nature of wireless channels by allowing users to relay information for each other so as to create multiple signal paths. In this talk I will discuss what is the best strategy from the viewpoint of a resource allocation protocol, to match source coding with cooperation diversity for conversational multimedia communications by studying the distortion performance for different schemes. I will present results that show that the best performance is obtained when all layers of a layered-coded source are sent with user cooperation (using decode-and-forward in most cases) if the source-destination channel is bad, and with no user cooperation, if the source-destination channel is good. The results also show that the gains from cooperative diversity outweigh the loss due to the sacrifice in overall bandwidth and that cooperation performance is sensitive to the proportion of communication capacity allocated for cooperation.
Date: 9/27/2005
Speaker: Thanongsak Himsoon
Title: Lifetime Maximization by Cooperative Sensor Protocol and Relay Deployment in Wireless Sensor Networks
Date: 9/20/2005
Speaker: Wipawee Siriwongpairat
Title: Employing Cooperative Diversity for Performance Enhancement in UWB Systems
Due to limitation on transmitted power level, any UWB system faces major design challenges in achieving wide coverage while assuring an adequate system performance. In this paper, an employment of cooperative communications in UWB is proposed to enhance UWB system performance by exploiting the broadcasting nature of wireless channels and the cooperation among UWB devices. Symbol-error-rate (SER) performance analysis and optimum power allocation are provided for cooperative UWB multiband OFDM systems with decode-and-forward cooperative protocol. To capture the multipath-clustering phenomenon of UWB channels, the SER performance is characterized in terms of cluster and ray arrival rates. An optimum power allocation is determined based on two different objectives, namely minimizing the overall transmitted power and maximizing the system coverage. Furthermore, an improved cooperative UWB multiband OFDM scheme is proposed to take advantage of unoccupied subbands. Simulation results are shown to validate the theoretical analysis.
Title: Bandwidth-Efficient OFDM Cooperative Protocol with Applications to UWB Systems
In this paper, we propose an OFDM cooperative protocol that not only achieves full diversity but also efficiently utilizes available bandwidth. The proposed protocol exploits limited feedback from the destination terminal (central node) such that each relay is able to help forward information of multiple sources in one OFDM symbol. To specify how relay-source pairs should be assigned, we propose two practical relay-assignment schemes, including fixed-location scheme in which the relays are fixed at optimum locations, and centralized-control scheme in which the relays are assigned by the central node. We provide outage probability analysis of the proposed protocol in wireless indoor environment. Moreover, a lower bound on the outage probability of any relay-assignment schemes is established, and the performanceof the proposed relay-assignment schemes is analyzed. We also investigate the application of the proposed protocol to enhance the performance of UWB systems. In UWB wireless indoor scenarios, both theoretical and simulation results show that the proposed cooperative protocol can achieve 75% power saving and 200% coverage extension compared to the non-cooperative UWB system proposed in the IEEE 802.15.3a standard.
Date: 9/13/2005
Speaker: Wei Yu
Title: Cooperation Enforcement and Security in Autonomous Ad Hoc Networks
In autonomous ad hoc networks, nodes belong to different authorities and pursue different goals, therefore cooperation among them cannot be taken for granted. On the contrary, to maximize their own benefits, nodes may tend to be selfish. Furthermore, some nodes may be malicious, whose objective is to cause damage to the network. In this paper we formally analyze the cooperation and security in autonomous ad hoc networks under a game theoretical framework. When analyzing the cooperation strategies, besides Nash equilibrium, other optimality criteria, such as Pareto optimality, subgame perfection, fairness, cheat-proofing, and noise resistance, have also been considered. We first study a simple yet illuminating two-player packet forwarding game, and conclude that a necessary condition for a strategy to be optimal from a selfish node's point of view is not to help its opponent more than its opponent has helped him. The obtained results are then extended to the multi-player packet forwarding game, both noiseless and noisy situations have been studied and the optimal strategies are proposed. Furthermore, malicious behaviors have also been investigated, and the optimal strategies are studied under different situations. The analysis shows that when the proposed strategies are used by all the selfish nodes in the network, the damage that can be caused by attackers is bounded as well as very limited. More surprisingly, our results show the presence of attackers may even help improving the network performance.

### Summer 2005

Date: 8/16/2005
Speaker: Zhu Ji
Title: Belief-Based Cooperation Enforcement for Autonomous Mobile Ad-hoc Networks with Noise and Imperfect Observation
In autonomous mobile ad hoc networks (MANET) where each user is its own authority, fully cooperative behaviors, such as unconditionally forwarding packets for each other, cannot be directly assumed. In this paper, we focus on cooperation enforcement in the autonomous mobile ad hoc networks with noise and imperfect observation and study the basic packet-forwarding function using the repeated game models with imperfect information. A belief-based packet forwarding framework is proposed to obtain cooperation-enforcement strategies solely based on each node's own past actions and his imperfect observation of other nodes' information. More importantly, the proposed strategy with belief system can not only achieve sequential equilibrium of the packet-forwarding game but also approach the optimal cooperation payoffs for two-player and multi-player cases. The simulation results illustrate that the proposed belief-based packet forwarding approach achieves near-optimal performance in the ad hoc networks with noise and imperfect observation.
Date: 8/9/2005
Speaker: Steve Tjoa
Title: Fingerprint Multicast for Scalable Video Coding
To reduce the bandwidth for watermark distribution, multicast is used as an alternative to unicast. We exploit the fact that, when employing spread-spectrum embedding upon a transformed image or video, there exist some coefficients that are more suited for watermark embedding than others. Only these coefficients should be watermarked and distributed separately. Unfortunately, this distribution scheme has many assumptions that include error-free transmission, spatial resolution, frame rate, etc. In the presence of a heterogeneous network where users require different levels of quality, we require a scalable video coding method. In this presentation, we will explore how watermarking can be used in conjunction with a scalable video codec. In particular, we discuss problems regarding video communication such as error protection and bandwidth efficiency, along with security issues such as watermark robustness and the embedding process.
Date: 8/2/2005
Speaker: Hong Zhao
Title: Multimedia Forensics for Traitor Tracing
Recent development in multimedia processing and network technologies has raised the critical issue of protecting multimedia content security and enforcing digital rights. Cryptographic tools (e.g., encryption and authentication) alone are not sufficient since the protection disappears after the data are delivered to users. To address the post-delivery protection, multimedia fingerprinting is an emerging technology to identify users who have legitimate access to the plaintext content but use it for unintended purposes. It provides a proactive means to trace the illegal usage of multimedia by inserting unique identification information ("fingerprint") into the content before distribution.

However, the uniqueness of each distributed copy poses new challenges to multimedia fingerprinting. The global nature of Internet enables a group of attackers to collectively mount attacks and effectively remove traces of the identifying fingerprints. Such a multi-user collusion attack poses serious threats to multimedia forensics. Digital fingerprints should resist multi-user collusion as well as attacks by a single adversary. Furthermore, for networked video applications with a large number of users, enabling traitor tracing and employing digital fingerprinting technology complicate the secure and efficient distribution of multimedia. This comes from the fact that traditional multicast technology cannot be directly applied to fingerprinted multimedia since different users receive slightly different copies. Therefore, in networked multimedia applications, it is critical to collectively address the security and the bandwidth efficiency and design secure fingerprint multicast schemes.

This talk addresses various issues in digital fingerprinting and discusses recent advances in traitor tracing for multimedia forensics. Traitor tracing using multimedia fingerprinting will be presented, including the investigation of multi-user collusion and collusion resistance of multimedia fingerprinting. In addition, the dynamics of the multimedia forensic system and behavior forensics in multimedia fingerprinting will be formulated and analyzed. Finally, secure fingerprint multicast schemes will be presented, which securely and efficiently distribute fingerprinted multimedia over networks.

Date: 7/26/2005
Speaker: Wei Yu
Title: Securing Cooperative Ad-hoc Networks Against Inside Attacks in a Game Theoretic Framework
In cooperative ad hoc networks, nodes usually belong to same authorities and pursue common goals. Since nodes in such networks will usually unconditionally help others, without necessary countermeasures, they are extremely vulnerable to inside attacks. In this paper we investigated how to secure cooperative ad hoc networks against inside attacks in a game theoretic framework. We focused on the most basic networking function, namely packet forwarding, and modeled the interactions between good nodes and inside attackers as secure packet forwarding games. The worst case scenario was studied where good nodes have no prior knowledge of the other nodes' types while inside attackers can know the types of all the other nodes. The Nash equilibrium strategies for the secure packet forwarding games under noiseless environment were first proposed. Then, the effects of noise on packet forwarding strategy design were investigated, and the Nash equilibrium strategies for the secure packet forwarding games under noisy environment were also proposed. Both analysis and simulation show that when the proposed strategies are adopted by good nodes, the damage that can be caused by the inside attackers is bounded as well as limited, and the performance of good nodes can be guaranteed.
Date: 7/19/2005
Speaker: Zhu Han
Title: Relay-Assignment Protocols for Coverage Extension in Cooperative Communications over Wireless Networks
One of the most relevant questions in implementing cooperative diversity protocols is how relay-source pairs should be assigned. In this paper, we address this problem and propose practical protocols for relay-assignment in cooperative wireless networks. We also investigate the application of the proposed protocols in coverage area extension in wireless networks. Outage probability is provided as a performance measure for the proposed protocols. First, we derive the performance of a hypothetical Genie-Aided protocol that provides an upper bound on the performance of any relay-assignment algorithm. This hypothetical protocol assumes that there is a Genie that, for each source node, places a relay at the optimal location to help that source. Then we describe and analyze the performance of two practical algorithms: the base-station/access-point (BS/AP) controlled protocol, and the Nearest-Neighbor protocol. The BS/AP-controlled protocol can be implemented at the BS or the AP, and it tries to emulate the Genie-Aided protocol. The second protocol, Nearest-neighbor, is a simple distributed protocol, in which the assigned relay is chosen to be the node between the user and the BS/AP such that it is the nearest neighbor to the user. We provide outage probability analysis for all the proposed protocols, and our analytical results are validated via computer experiments. Simulation results for two communication setups are considered: rural scenarios for cellular systems, and indoor scenarios for wireless local area networks (WLAN). By utilizing the proposed protocols, simulation results indicate a signicant gain in coverage area over the direct transmission scheme. For the rural cellular system case, coverage extension of about 250% can be achieved by the BS/AP-controlled protocol. For the indoor WLAN, a 350% increase in the coverage area can be achieved by both the BS/AP-controlled and Nearest-neighbor protocols, as both protocols provide comparable performance in the indoor WLAN scenario.
Date: 7/19/2005
Speaker: Prof. Lee Swindlehurst, Brigham Young University
Title: Some Interesting Research Problems in Communications and Signal Processing
This talk will briefly outline several open problems that we are addressing at BYU, in an effort to find areas of common interest with CSPL. The topics to be discussed include: (1) time-of-arrival estimation for MIMO-OFDM channel estimation, (2) semi-blind uplink source separation using constant-modulus signals, (3) sensor deployment for source localization and tracking, (4) resource allocation for downlink spatial multiplexing, (5) MIMO channel prediction, and (6) scheduling in multi-antenna ad hoc networks.
Date: 7/12/2005
Speaker: Beibei Wang
Title: Scheduling Algorithms in Wireless Networks
The design objective in wireless network scheduling algorithms is to guarantee Quality of Service (QoS) for each mobile user in a resource-shared environment and meanwhile maximize the total system throughput. Based on the Effective Capacity (EC) link model, we introduced a notion of delay-bound violation/tolerance probability as a QoS measure. Then Maximum Utility Scheduling (MUS) scheme is proposed, which jointly considers throughput maximization and QoS support. Basically, the approach has two major steps: First, select a user that has the maximum utility. Second, recursively estimate the EC link model functions and update corresponding QoS parameters in the utility function. Simulation results demonstrate that even under heavy network loads our proposed scheme can tolerate delay of service with a higher probability than other scheduling algorithms and achieve high throughput. Moreover, system overall performance can be optimized by properly choosing some parameter.
Date: 7/5/2005
Speaker: Wan-Yi Lin
Title: Detection Statistics of Multi-user Collusion on Scalable Fingerprinting Systems
We all know that due to some communication and computation constraint, the multimedia offers different resolution. So the fingerpringting should also have this scalibility to handle this situation. And during collusion, the colluders may have different resolution copies, to achieve fairness, the colluders assume that the detector detect based on the whole sequence, but what will happen if this assumption fail? In today's meeting, I'll present the statistics if the detecor detect layer-by-layer, and what will the colluders do to attain their fairness.
Date: 6/28/2005
Speaker: Peng Qiu
Title: Evolutionary Programming with Application in Medical Imaging
We are already familiar with some optimization methods such as steepest decent, Newton methods, conjugate gradient, and so on. These methods all belong to single search strategy. Basically, we stick to one guess (potential solution) and try to improve it iteratively. However, these method may not work well when there are multiple local solutions. Evolutionary programming (EP) adpots a different strategy. It works with multiple "guesses". In each iteration of EP, there are a population of potential solutions. And these potential solutions interact in a way that mimics the nature (evolution, animal's social system). The basic idea is to "let the nature find its way". EP is a stochastic strategy that ENCOURAGES all potential solutions to move toward better places. And this kind of methods is able to deal with problems with multiple local optimals. After introducing EP, I'll talk about its application in PET imaging. In PET imaging treatment, the patient's blood sample is tested from time to time. This measurement is called "input function", which it quite painful and risky. However, we found that this measurement is not necessary, and we are able to estimate it. In this talk, I will formulate the estimation as a minimization problem, and try to solve it with an evolutionary programming method called Particle Swarm Optimation. And let's see how intelligent swarms could be.
Date: 6/21/2005
Speaker: Chaiyod Pirak
Title: Cooperative Communications
In the future wireless communications, high speed data transmission services are highly demanded. Several schemes have recently been proposed for realizing such demands, such as Space-Time (ST) coded MIMO systems and Space- Frequency (SF) coded MIMO-OFDM systems. However, by given a current technology of wireless communications, equipping multiple antennas at the receiver or transmitter seems to be far from realistic. One mean of implementing the MIMO systems is to employ a concept of distributed antenna systems, namely cooperative communications. In this talk, I will present an efficient protocol for cooperative communications, namely a minimum SER selection adaptive protocol that exploits an upper bound on SER as a performance-comparing threshold for optimally choosing a modulation constellation and cooperative partners. Our proposed protocol also employs an adaptive modulation for maintaining a fixed bandwidth efficiency constraint. In addition, an optimum power allocation is also addressed in this study. Extensive computer simulations for evaluating the performance of the proposed protocol are provided. By given the fixed bandwidth efficiency and power constraints, our proposed protocol performs the best in comparison to such fixed protocols. Furthermore, it has been shown in the simulation results that the optimum power allocation scheme is superior to the equal power allocation scheme, in which the SNR difference is about 0.65 dB at SER of 10^-5. Although the proposed protocol is suboptimal in low SNR regimes, it has also been shown in the simulation results that it works remarkably well in all SNR levels, where it achieves the minimum SER through out the SNR range.

### Spring 2005

Date: 5/18/2005
Speaker: Chaiyod Pirak
Title: Space-Frequency Coded MIMO-OFDM Channel Estimation
Multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) is one of a prominent communication system for realizing high speed data transmission services. One critical issue for such systems employing coherent receivers is channel estimation. Since the multipath delay profile of channels is arbitrary in the MIMO-OFDM systems, an effective channel estimator is needed to estimate such channels. In this paper, we first develop a pilot-embedded data-bearing approach for joint channel estimation and data detection, in which pilot-embedded data-bearing (PEDB) least-square (LS) channel estimation and maximum-likelihood (ML) data detection are employed. Then we propose an LS FFT-based channel estimator by employing the concept of FFT-based channel estimation to improve the PEDB- LS channel estimation via choosing certain significant taps in constructing a channel frequency response. The effects of model mismatch error inherently in the proposed LS FFT-based channel estimator when considering non-integer multipath delay profiles, and its performance analysis are investigated. Under the framework of pilot embedding, we further propose an adaptive LS FFT-based channel estimator that employs the optimum number of taps such that an average total energy of the channels dissipating in each tap is completely captured in order to compensate the model mismatch error as well as minimize the corresponding noise effect to improve the performance of the LS FFT-based channel estimator. Simulation results reveal that the adaptive LS FFT-based channel estimator is superior to the LS FFT-based and PEDB-LS channel estimators under quasi-static channels or low Doppler's shift regimes.
Date: 5/4/2005
Speaker: Zhu Ji
Title: An Optimal Dynamic Pricing Framework for Autonomous Mobile Ad-hoc Networks
In autonomous mobile ad hoc networks (MANET) where each user is its own authority, fully cooperative behaviors, such as unconditionally forwarding packets for each other or, honestly revealing its private information, cannot be directly assumed. The pricing mechanism is one way to provide incentives for the users to act cooperatively by awarding some payment for cooperative behaviors. In this talk, we model the pricing and routing as multi-stage dynamic games. By taking into consideration that the packet-forwarding will incur a cost to the relay user and the successful transmission brings benefits to the sender/receiver, we propose a dynamic pricing framework to maximize the sender/receiver's payoff by considering the dynamic nature of MANETs, meanwhile, keeping the forwarding incentives of the relay nodes by providing the optimal payments based on the auction rules. The contributions of this paper are multi-folds: Firstly, by modeling the pricing and routing as a dynamic game, the sender is able to exploit the time diversity in MANET to increase their payoffs by adaptively allocating the packets to be transmitted into different stages. Secondly, based on the auction structure and routing dynamics, a simple optimal dynamic programming algorithm is developed to implement efficient multi-stage pricing for autonomous MANETs. Thirdly, the path diversity of MANET is exploited using the optimal auction mechanism in each stage. The simulation results illustrate that the proposed dynamic pricing framework has significant performance gains over the static pricing algorithms.
Date: 4/27/2005
Speaker: Ahmed Ibrahim
Title: Cooperative Communication with Partial Channel State Information
In this talk, we propose a new cooperative protocol, which takes into consideration the channel state information (CSI) available at the source. With such protocol, a significant improvement in the transmission rate can be achieved in decode-and-forward cooperative transmission. Closed-form expressions for the transmission rate and the symbol error rate (SER) are derived for both M-PSK and M-QAM signalling. Moreover, two optimization metrics are considered in the protocol design to enhance the system performance, the first is based on minimizing the SER alone, while the second is based on minimizing a joint function of both the SER and the transmission rate. Finally, the obtained analytical results are verified through computer simulations.
Date: 4/20/2005
Speaker: Quoc Lai
Title: Ultra-wideband Communication
Ultra-wideband (UWB) has emerged as a technology that offers great promises to satisfy the growing demand for low cost and high-speed digital wireless home networks. In our group meeting today, I would like to present my work on the baseband implementation of UWB system employing multiband OFDM that has been proposed in the IEEE 802.15.3a standard. Basically, the transceiver comprises of data scrambler, convolutional en/decoder, bit-interleaver, constellation mapping, and IFFT/FFT. In the talk, I will describe these components in detail.
Date: 4/6/2005
Speaker: Karim Seddik
Title: Space-Time Collaborative Communications: Performance Analysis and Optimal Power Allocation; Multi-node Amplify and Forward Collaborative Networks
Symbol-error-rate (SER) performance analysis is provided for the decode-and-forward space time block coded cooperation protocol. We derive the SER for the case of two relays forwarding the source data and emulating the Alamouti scheme. The analysis presented can be easily generalized to any number of relays emulating orthogonal space-time block code. We derive closed-form SER expressions for both the M-PSK and M-QAM signals. Moreover, a SER bound is established which is tight at high signal-to-noise ratio. Based on this tight bound, we also determine the optimum power allocation for this space-time block coded cooperative network. Simulations are used to validate the theoretical results. Next, we will present two schemes for the multi-node amplify and forward collaborative networks. In the first scheme, each relay only amplifies the signal it receives from the source. We denote this system as the 2 phase system. In the second scheme, each relay applies maximum ratio combining (MRC) on the signals from the source and all previous relays. We denote this system as the MRC based system. We provide an SER bound for the 2 phase system. The analysis of the MRC system is intractable so we prove via simulations that it does not offer any improvement over the 2 phase system due to the noise propagation problem. We then provide a Lower bound on the SER performance of any amplify and forward system. We then prove that the 2 phase system achieves this bound if the relays are close to the source.
Date: 3/30/2005
Speaker: Charles Pandana
Title: Energy-Aware Packet Forwarding Optimization using Learning and Repeated Game Framework
In wireless ad hoc networks, autonomous nodes would be reluctant to forward others' packets because of the nodes' limited energy. However, such selfishness would deteriorate both the system efficiency and users' performances. Hence, it is crucial to design a distributed mechanism for enforcing the cooperation among greedy users for packet forwarding. Moreover, having only with the local information, the optimal cooperation is not known to distributed users/nodes, even though they are willing to cooperate. In this paper, we proposed a self-learning repeated game framework to study how to cooperate and maintain the cooperation among selfish nodes. The framework has two major schemes: First, an adaptive repeated game scheme ensures the cooperation among users for the current cooperative packet forwarding probabilities. The repeated game scheme provides the users with a mechanism that any deviating user would be punished enough by others in the future, so that no user has incentive to deviate. Second, a self-learning scheme tries to find the better cooperation probabilities that are feasible and benefit all users. Starting from noncooperation, the above two proposed schemes are applied iteratively, so that better cooperation is discovered and maintained in each iteration. From the simulation results, the proposed framework achieves solutions within up to 4% compared to the solution of a centralized system. Moreover, the proposed algorithm is able to enforce cooperation among selfish nodes.
Date: 3/2/2005
Speaker: Thanongsak Himsoon
Title: Differential Transmission for Amplify-and-Forward Cooperative Communications
We propose a differential transmission scheme for amplify-and-forward protocol in a two-user cooperative communications system. By efficiently combining signals from both direct and relay links, the proposed scheme provides superior performance to those of direct transmissions without relay in either differential detection or coherent detection. While the exact BER formulation of the proposed scheme is not available currently, we provide, as a performance benchmark, an exact bit error rate (BER) formulation for a case of optimumcombining cooperation system with M-ary differential phase shift keying signals. We also provide BER upper bound and lower bound as well as their simple approximations, in which one of them is tight at high signal-to-noise ratio (SNR). In addition, the optimum power allocation for the cooperation system is determined based on the provided BER formulations. We show that the proposed differential cooperative transmission scheme together with the optimum power allocation yields comparable performance to the optimum-combining scheme. Simulation result show that a significant performance improvement is obtained for a case of optimum power allocation strategy comparing to the performance with equal power allocation scheme.
Date: 2/23/2005
Speaker: Wipawee Siriwongpairat
Title: Performance Characterization of Multiband UWB Communication Systems using Poisson Cluster Arriving Fading Paths
In this talk, we provide performance analysis for UWB systems that successfully captures the unique multipath-rich property and multipath-clustering phenomenon of UWB channels. Using the Saleh-Valenzuela model, we characterize pairwise error probability (PEP) and outage probability for UWB systems employing multiband OFDM based on the cluster arrival rate, the ray arrival rate within a cluster, and the cluster and ray decay factors. Furthermore, an approximation technique is established, which allows us to obtain closed-form performance formulations that provide insightful understanding of the effect of channel characteristics on the performances of UWB systems. Finally, we characterize the effect of random-clustering phenomenon on the performance of UWB-MIMO systems. The theoretical results reveal that regardless of the clustering behavior of UWB channels, the diversity gain can be improved by increasing the number of jointly encoded subcarriers, the number of jointly encoded OFDM symbols, or the number of antennas. The coding gain on the other hand, depends heavily on the cluster-arriving channels. Extensive simulation results are provided to support the theoretical analysis.
Date: 2/16/2005
Speaker: Andres Kwasinski
Title: Joint Source Coding Diversity and Cooperative Diversity for Multimedia Communications
Source coding diversity produces multiple independent source descriptions so as to improve the received quality. Cooperative diversity exploits the broadcast nature of wireless networks by allowing multiple users to relay information for each other, so as to create multiple signal paths. In this talk, I will discuss what is the best strategy in combining these two types of diversity for multimedia communications. We study the distortion-rate performance for different joint diversity schemes. Our results show that exploring both types of diversity improves the system performance most, but in most of cases, only one type of diversity can achieve most of the diversity gain. The best performance is obtained when the mobile can switch between cooperative and non-cooperative operation depending on the channel conditions. Thus, it is important to design efficient protocols that manage this switch and provides motivation for all users to cooperate.
Date: 2/9/2005
Speaker: Hong Zhao
Title: Traitors Within Traitors in Digital Fingerprinting For Multimedia Forensics
Digital fingerprinting can be used to identify the source of illicit copies of multimedia data and enforce digital rights. Collusion involves multiple adversaries to collectively mount attacks against digital fingerprints and is a cost-effective method to undermine the traitor tracing capability. In this talk, we study the dynamics among attackers during collusion and consider the problem of traitors within traitors, in which some selfish colluders wish to minimize their own risk of being caught while still profiting from collusion. In particular, we investigate the strategy that selfish colluders can use to minimize their probability of being detected and analyze its performance.
Date: 2/2/2005
Speaker: Zhu Han
Title: How to Cooperate in Wireless Networks
Wireless networking and resource allocation are general strategies to utilize the limited wireless radio resources, control the co-channel interferences, and enhance the network performances. To reduce the overhead imposed by such strategies, the mobiles of the next generation networks have their own autonomies for resource allocation in a distributive way. However, the non-cooperative competition of the radio resources results in low system efficiency. So how to ensure cooperation among autonomous users is one of the most important wireless networking research topics. There are many types of approaches to enforcing cooperation such as incentive based, referee based, punishment based, collaborative communication based, etc. In this talk, we concentrate on the first two approaches and employ them to OFDMA networks.

For incentive based approach, we propose bargaining method to have mutual benefits. Specifically, a fair and simple scheme to allocate subcarrier, rate, and power for multiuser single cell OFDMA systems is considered. The problem is to maximize the overall system rate, under each user's maximal power and minimal rate constraints, while considering the fairness among users. The approach proposes the fairness and low complexity implementation based on Nash Bargaining Solutions and Coalitions. First, a two-user algorithm is developed to bargain subcarrier usage between both users. Based on this algorithm, we develop a multiuser bargaining algorithm where optimal coalition pairs among users are constructed. The simulation results show that the proposed algorithms not only provide fair resource allocation among users, but also have comparable overall system rate with the scheme maximizing the total rate without considering fairness. They also have much higher rates than those of the scheme with max-min fairness.

For referee based approach, we propose non-cooperative game approach with a referee for multi-cell OFDMA networks. The goal is to minimize the overall transmitted power under the constraints that each user has the desired throughput and each user's power is bounded. The pure non-cooperative game may have some undesired Nash Equilibriums with low performances. To improve the performances, a referee is introduced to the networks and is in charge of monitoring the outcome of non-cooperative competition of resources for the distributed users. If the outcome is not desired, some users should be kicked out from using the resources such as sub-channels or the required transmission rates should be reduced, so that the rest of users can share the resources more efficiently. From the simulation results, the proposed scheme reduces the overall transmitted power greatly compared with the fixed channel assignment algorithm and pure water-filling algorithm.

Collaborative communication based approach and punishment based approach are also briefly discussed to give an overview on how to cooperate in wireless networks.

Date: 1/26/2005
Speaker: Wei Yu