学术报告系列：机器学习、无线通信及信号处理
第一场
Title: Advances in Machine Learning Techniques with Applications to Communications and Image Processing
Speaker: Prof. Yik-Chung Wu, The University of Hong Kong
Abstract: This talk shall briefly overview the recent advances of machine learning techniques at the University of Hong Kong, and how they are applied to communications and image processing problems. The specific topics to be introduced include Gaussian belief propagation, variational statistical inference and tensor processing. Surprisingly, many machine learning models are general enough to tackle problems that are seemingly unrelated. The applications to be covered include synchronization in large-scale networks, channel estimation under high mobility, power state estimation in smart grid, massive MIMO channel estimation, face classification, surveillance video objects separation, and image de-noising.
Bio: Yik-Chung Wu received the B.Eng. (EEE) degree in 1998 and the M.Phil. degree in 2001 from the University of Hong Kong (HKU). He received the Croucher Foundation scholarship in 2002 to study Ph.D. degree at Texas A&M University, College Station, and graduated in 2005. From August 2005 to August 2006, he was with the Thomson Corporate Research, Princeton, NJ, as a Member of Technical Staff. Since September 2006, he has been with HKU, currently as an Associate Professor. He was a visiting scholar at Princeton University, in the summers of 2011 and 2015. His research interests are in general area of signal processing, machine learning and communication systems, and in particular distributed signal processing and communications; and large-scale and robust optimization. Dr. Wu served as an Editor for IEEE Communications Letters, is currently an Editor for IEEE Transactions on Communications and Journal of Communications and Networks.
第二场
Title: Impact of antenna correlation on full-duplex two-way massive MIMO relaying systems
Speaker: Prof. Shaodan Ma, University of Macau.
Abstract: Full duplex and massive multiple input multiple output (MIMO) have been regarded as promising techniques for 5G communications. When a massive number of antennas are deployed, antenna correlation is usually unavoidable due to limited spacing among antennas. This talk will investigate the impact of antenna correlation on full duplex massive MIMO two-way relaying systems. Deterministic equivalent of the achievable sum-rate is first given, which enables the derivation of the asymptotic sum-rate when the number of relay antennas is large. The impact of antenna correlation on the asymptotic sum-rate is then thoroughly analyzed. The results reveal that with a large number of relay antennas, the asymptotic sum rate is independent of the antenna correlation at the relay’s transmitter side, however it is Schur-concave with respect to the eigenvalue vectors of the involved antenna correlation matrices at both the users’ transceiver sides and the relay’s receiver side. In other words, the antenna correlations at both the users’ transceiver sides and the relay’s receiver side have detrimental effects on the sum-rate. Moreover, the antenna correlation at the users is more significant to the asymptotic sum-rate than that at the relay. The analysis is general and can be easily reduced to consider half-duplex and/or one-way massive MIMO relaying systems. Numerical results are finally provided to validate the analysis. It is also shown that a larger number of user antennas may not always lead to a higher sum-rate, and thus proper selection of the number of user antennas is necessary to maximize the sum-rate.
Bio: Shaodan Ma received her double Bachelor degrees in Science and Economics, and her Master degree of Engineering, from Nankai University, Tianjin, China. She obtained her Ph. D. degree in electrical and electronic engineering from the University of Hong Kong, Hong Kong, in 2006. After graduation, she joined the University of Hong Kong as a Postdoctoral Fellow. Since August 2011, she has been with the University of Macau and is now an Associate Professor there. She was a visiting scholar in Princeton University in 2010 and is currently an Honorary Assistant Professor in the University of Hong Kong. She was a symposium co-chair for Signal Processing Symposium in the Ninth International Conference on Wireless Communications and Signal Processing (WCSP 2017), Wireless Communications Symposium in 2016 IEEE Global Communications Conference (GlobeCom 2016), Signal Processing for Communications Symposium in 2016 IEEE International Conference on Communications (ICC 2016), Track on Fundamentals and PHY in 2015 IEEE International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC 2015). She serves as an editor for IEEE Wireless Communications letters since 2017. She was a co-recipient of Best Paper Awards in the 2012 International Conference on Wireless Communications and Signal Processing (WCSP 2012) and the 2014 IEEE International Conference on Communication Systems (ICCS 2014). Her research interests are in the general areas of signal processing and communications, particularly, transceiver design, resource allocation and performance analysis.
第三场
Title: Optimal Training Design for MIMO Systems with General Power Constraints
Speaker: Prof. Chengwen Xing, Beijing Institute of Technology
Abstract: Training design for general multiple-input multiple output (MIMO) systems is investigated in this paper. Unlike prior designs which are applicable only for centralized MIMO systems with total power constraint, general power constraints are considered here. They cover total power constraint and individual power constraints as special cases and are thus applicable for not only centralized but also distributed MIMO systems. By writing the MIMO received signals in matrix and vector forms respectively, three channel estimation schemes, i.e., right estimation, left estimation, and right-left estimation, are discussed. Their corresponding training designs are tackled individually with the general power constraints. Under each channel estimation scheme, optimal training sequences to maximize the mutual information between the true channel and its estimate and minimize the mean square error (MSE) of the channel estimate are respectively proposed in closed-forms. The relationship between the two design criteria, i.e., mutual information maximization and MSE minimization, are clearly revealed. The optimal training designs under the three estimation schemes are also compared in depth. It is found that the right estimation exploits less statistical information of the channel and noise, and provides worse performance than the left estimation but with lower computational complexity. On the other hand, the right-left estimation performs in between and offers better tradeoff between the performance and computational complexity than the other two estimations. The optimality and effectiveness of the proposed training designs are finally verified by extensive simulations.
Bio: Chengwen Xing received the B.Eng. degree from Xidian University, Xian, China, in 2005 and the Ph.D. degree from the University of Hong Kong, Hong Kong, in 2010. Since September 2010, he has been with the School of Information and Electronics, Beijing Institute of Technology, Beijing, China. From September 2012 to December 2012, he was a Visiting Scholar with the University of Macau, Macau, China. His current research interests include statistical signal processing, convex optimization, multivariate statistics, combinatorial optimization, massive multiple-input multiple-output systems, and high-frequency band communication systems. Dr. Xing is an Associate Editor of the IEEE TRANSACTIONS ON VEHICULAR TECHNOLOGY, the KSII Transactions on Internet and Information Systems, the Transactions on Emerging Telecommunications Technologies, and China Communications.
第四场
Title: Performance Analysis and Optimal Design of HARQ Systems Over Time-Correlated Fading Channels
Speaker: Dr. Zheng Shi, Jinan University
Abstract: Outage performance is first analyzed for three types of HARQ systems over time-correlated fading channels, including Type I Hybrid Automatic Repeat Request (HARQ), HARQ with Chase Combining (HARQ-CC) and HARQ with Incremental Redundancy (HARQ-IR). Unlike prior analyses, time-correlated Nakagami-$m$ fading channel is considered. The outage analysis thus involves the probability distribution analysis of multiple correlated Gamma random variables and is more challenging than prior analyses. Different analytical methods are applied to derive the outage probabilities of three types of HARQ schemes. The analytical results enable asymptotic outage analysis to extract clear insights, which have never been discovered for HARQ systems even under fast fading channels. The asymptotic outage probability is then derived in a simple form which clearly quantifies the impacts of transmit powers, channel time correlation and information transmission rate. It is proved that the asymptotic outage probability is an inverse power function of the product of transmission powers in all HARQ rounds, an increasing function of the channel time correlation coefficients, and a monotonically increasing and convex function of information transmission rate. The simple expression of the asymptotic result enables optimal power allocation and optimal rate selection of HARQ with low complexity under various system design objectives. Finally, some potential research directions are briefly introduced to further extend our prior work.
Bio: Zheng Shi received the B.S. degree in communication engineering from Anhui Normal University, China, in 2010 and the M.S. degree in communication and information system from Nanjing University of Posts and Telecommunications (NUPT), China, in 2013. He obtained his Ph.D. degree in electrical and computer engineering from University of Macau, Macao, in Jul. 2017. From Sept. 2016 to Feb. 2017, he was a visiting student in King Abdullah University of Science and Technology. Since Sept. 2017, he has been a lecture with the School of Electrical and Information Engineering, Jinan University. His research interests include performance analysis of wireless communication systems, HARQ, NOMA, millimeter wave and heterogeneous wireless networks.
第五场
Title: Advances in Energy Harvesting: Nonlinear Model and Vehicular Technology
Speaker: Mr. Shuai Wang, The University of Hong Kong
Abstract: Energy harvesting is an enabling technique for Internet of Things. While the theory of wireless energy harvesting has been well studied recently, there is a significant gap between the theoretic result and the reality. In particular, the harvested direct-current power is currently modelled as a linear function of the incident radio-frequency power. However, since the energy harvester contains nonlinear elements such as diodes, linear energy harvesting models cannot accurately reflect the energy conversion process. To capture the dynamics of energy harvesting circuits, this talk presents a novel nonlinear energy harvesting model which can match the Powercast energy harvesting product P2110 very well. On the other hand, it is well-known that a major obstacle to wireless energy harvesting is the path-loss. By endowing the power beacon with mobility, the distances between the power beacon and users can be varied, thus providing a potential solution to combat path-loss at the expense of energy for transmission. By comparing the performance for the fixed power beacon case and the mobile power beacon case, it is possible to quantify the relative advantage of spending energy n moving versus on transmission.
Bio: Shuai Wang received his B.S and M.S. degrees from Beijing University of Posts and Telecommunications in 2011 and 2014, respectively, both in Electronic Engineering. Currently, he is pursuing Ph.D. degree in Electrical and Electronic Engineering, at the University of Hong Kong. His research interests include signal processing and analysis of wireless networks using optimization.
第六场
Title: Belief Propagation: convergence analysis and applications
Speaker: Mr. Bin Li, The University of Hong Kong
Abstract: In many probabilistic inference problems, computing marginal distributions from joint distribution is an essential task. Brute-force method by taking summations or integrals directly can be costly and inefficient, especially for joint distribution with large numbers of variables. Belief propagation (BP) is a low-complexity and efficient message passing algorithm for performing inference on graphical models. It has been widely used to compute marginal distributions from joint distribution. However, BP involves the convergence problem. For acyclic graphs, BP is guaranteed to obtain the exact marginal distributions. However, for loopy graphs, BP may diverge. Fortunately, if BP does converge in loopy graphs, it obtains the exact means of marginal distributions. Therefore, understanding the convergence of BP is important to guarantee the performance in probabilistic inference problems. Besides, BP has the advantages over paralleled and distributed computation, and is widely used in wireless communication, signal processing, and machine learning.
Bio: Bin Li received his B.Eng. degree in 2012 and M. Eng. degree in 2015 from Beijing Institute of Technology (BIT). He is currently working toward the Ph.D. degree at the University of Hong Kong (HKU). His research interests are signal processing, machine learning, and wireless communications.
第七场
Title: Robust Energy-Efficient Precoding Optimization for Dual-Polarized Multiuser MIMO
Speaker: Ms. Gui Zhou, Beijing Institute of Technology
Abstract: We investigate the robust fairness-based energy efficiency (EE) precoding optimization of the dual-polarized multiuser multiple-input multiple-output (MIMO) downlink. Exploiting the dual-polarized antenna structure, the dual-structured linear precoding scheme is adopted for achieving low implementation complexity, low feedback overhead and good performance. The polarization-based subgrouping technique is also utilized to divide spatially grouped users into co-polarized subgroups to further reduce channel feedback overhead. The proposed robust fairness-based EE precoding design is capable of maximizing the minimum EE, i.e., the worst-case EE, achieved among all users with the norm-bounded channel errors. This robust EE optimization is naturally decomposed into two stages. In the first stage, based on the polarized spatial correlation information, the preprocessing matrix is optimized via the block diagonalization. In the second stage, the linear precoding matrix is optimized by exploiting the equivalent relationship between the signal-to-interference-plus-noise ratio and the mean square error, based on the fractional programming method and the sign-definiteness lemma. Specifically, the corresponding nonconvex EE fractional optimization problem can be transformed into a series of semidefinite programming problems, which are efficiently solved by the convex optimization technique. Simulation results are included to demonstrate the robust EE performance advantages of the proposed dual-structured precoding for the dual-polarized multiuser MIMO downlink.
Bio: Gui Zhou received her B. Eng. degree in 2015 from Beijing Institute of Technology (BIT). She is currently working toward the Ph.D. degree with the School of Electronic and Information, Beijing Institute of Technology. Her research interests are array signal processing, and wireless communications.
第八场
Title: Coordinated Multi-Point Transmission: A Poisson-Delaunay Triangulation Approach
Speaker: Ms. Yan Li, Sun Yat-sen University
Abstract: In my work, a novel analytical model for coordinated multi-point (CoMP) operation in cellular networks is proposed, where we analyze three Types of user equipment (UE) and two different joint processing mechanisms, namely, joint transmission and dynamic point selection/muting are employed at the BSs for the worst-case UEs. Our results show that the Poisson-Delaunay triangulation based approach can significantly improve the quality-of-service of the users in cellular networks, especially for the worst-case users. Unlike conventional coordinated multi-point transmission techniques in Poisson-Voronoi tessellated networks where the cooperating BS set of a particular user is dynamic and needs on-line updated occasionally, the cooperating BS set of a user in Poisson-Delaunay triangular cells is fixed and can be off-line determined according to the geometric locations of BSs. Moreover, the coverage area of the network is tessellated based on the Poisson-Delaunay triangulation, yielding shrinkage of the average cell size to half, although at the cost of cooperating gains.
Bio: Yan Li obtained her B.S. degrees in Electronic Information Engineering from Hunan Normal University, Changsha, China, in 2013 and M.S. degree from Sun Yat-sen University, Guangzhou, China, in 2016. She is currently working towards the Ph.D. degree at the School of Information and communication engineering, Sun Yat-sen University, Guangzhou, China. Her current research interest is coordinated multi-point (CoMP) transmission for wireless networks.