Tutorials

ICASSP 2014, along the tradition of previous ICASSPs, will house 15 high quality tutorials on hot topics for the signal processing community. This year, we are providing a rich tutorial list covering diverse areas of signal processing. The aim of a tutorial is to provide an overview of the state-of-the-art in a specific field of signal processing.

Each tutorial will be three-hour long with one, two or three speakers. Tutorials are scheduled before the start of regular conference sessions. Five tutorials will be provided in parallel sessions during three time slots on Sunday PM, Monday AM, and Monday PM. Care has been taken not to overlap tutorials from the same subject areas.
Supporting material for the tutorial will be provided to the attendees.

Ercan Kuruoglu and Antonio Napolitano,
ICASSP 2014 Tutorial Chairs

List of Tutorials

SUNDAY, MAY 4, 2014, AFTERNOON
14.00 - 17:30 (30 minutes coffee break at 15.30)

T1 - Statistical Signal Processing for Graphs
Subject Area: Fundamentals
Speakers: Nadya T. Bliss (Arizona State University), Alfred O. Hero (University of Michigan, Ann Arbor), Benjamin A. Miller (MIT Lincoln Laboratory)

T2 - Monotone Operator Splitting Methods in Signal and Image Recovery
Subject Area: Image Processing
Speakers: P.L. Combettes (Université Pierre et Marie Curie – Paris 6), J.-C. Pesquet (Université Paris-Est), and N. Pustelnik (ENS de Lyon)

T3 - Informed Audio Source Separation: Trends, Approaches and Algorithms
Subject Area: Speech/Audio/Language Processing
Speaker: Alexey Ozerov (Technicolor), Antoine Liutkus (INRIA, Nancy Grand Est) and Gaël Richard (Telecom ParisTech)

T4 - Signal Processing for Analog Systems
Subject Area: Signal Processing System Design and Implementation
Speakers: Arthur J. Redfern, Manar El-Chammas and Lei Ding (Texas Instruments)

T5 - Transmitter Cooperation in Wireless Networks: Potential and Challenges
Subject Area: Communications
Speakers: David Gesbert and Paul de Kerret (EURECOM)


MONDAY, MAY 5, 2014, MORNING
9.30 - 13.00 (30 minutes coffee break at 11.00)


T6 - Signal Processing for Big Data
Subject Area: Fundamentals
Speakers: G.B. Giannakis, Konstantinos Slavakis (University of Minnesota), Gonzalo Mateos (Carnegie Mellon University)

T7 - Semidefinite Relaxation: From Theory to Applications to Latest Advances
Subject Area: Fundamentals
Speakers: Wing-Kin Ma and Anthony Man-Cho So (The Chinese University of Hong Kong)

T8 - EEG Signal Processing and Classification for Brain Computer Interfacing (BCI) Applications
Subject Area: Biomedical signal processing
Speakers: Amit Konar (Jadavpur University), Fabien Lotte (INRIA-Bordeaux Sud-Ouest), Arijit Sinharay (Tata Consultancy Services Ltd)

T9 - Deep learning for natural language processing and related applications
Subject Area: Speech/Audio/Language Processing
Speakers: Xiaodong He, Jianfeng Gao, Li Deng (Microsoft Research)

T10 - Bits and Flops in modern communications: analyzing complexity as the missing piece of the wireless-communication puzzle
Subject Area: Communications
Speakers: Petros Elia (EURECOM) and Joakim Jaldén (Royal Institute of Technology, KTH, Sweden)


MONDAY, MAY 5, 2014, AFTERNOON
14.00 - 17:30 (30 minutes coffee break at 15.30)

T11 - An introduction to sparse stochastic processes
Subject Area: Fundamentals
Speaker: Micheal Unser (EPFL)

T12 - Factoring Tensors in the Cloud: A Tutorial on Big Tensor Data Analytics
Subject Area: Fundamentals
Speakers: Nicholas Sidiropoulos (University of Minnesota) and Evangelos Papalexakis (Carnegie Mellon University)

T13 - Complex elliptically symmetric distributions and their applications in signal processing
Subject Area: Statistical Signal Processing
Speakers: Esa Ollila (Aalto University, Finland), David E. Tyler (Rutgers University) and Frederic Pascal (SUPELEC)

T14 - Signal Processing for Finance, Economics and Marketing Modeling and Information Processing
Subject Area: Financial data analysis
Speakers: Xiao‐Ping (Steven) Zhang (Ryerson University), Fang Wang (Wilfrid Laurier University)

T15 - Signal Processing in Power Line Communication Systems
Subject Area: Communications
Speaker: Andrea M. Tonello (University of Udine, Italy)

Abstracts of Tutorials

T1 - Statistical Signal Processing for Graphs
Subject Area: Fundamentals
Speakers: Nadya T. Bliss (Arizona State University), Alfred O. Hero (University of Michigan, Ann Arbor), Benjamin A. Miller (MIT Lincoln Laboratory)

Summary
In numerous applications, detection of anomalous activity among a small subset of nodes in a network is a problem of significant interest. Casting this problem in the context of traditional signal detection has the potential to provide a framework enabling analysis of the detectability of small anomalies in large networks. However, the combinatorial nature of graphs—the common mathematical object for network representation—complicates the application of detection theory, requiring NP--‐hard problems to be solved for optimal detection. Using classical signal processing techniques of regression and residuals analysis, leads to tractable algorithms for anomaly detection in large, noisy, dynamic graphs. This tutorial will focus on solid fundamentals of signal processing techniques applied to graphs and networks as they arise in emerging signal processing applications (social media, information networks, biological networks, etc.). Common properties of real datasets will be presented, and the tutorial will provide an introduction to random graph models that incorporate these common features. Additionally, examples of application of these general techniques to a broad range of datasets will be covered, and the tutorial will also impart to attendees a concrete sense of the variety of network data types currently in wide use. Examples highlighted in the tutorial will be drawn from different application areas, such as information forensics, social network analysis, power networks, and biological and biomedical networks.

Outline
I. Background and Tools:
A. Example applications
B. Classical graph tools
C. Graph algorithms in computer science
D. Sparse matrix algebra
E. Spectral graph theory
II. Models and Data
A. Empirical properties of real data
B. Network centrality measures
C. Basics of random graphs
D. Signal and noise models, filtering
III. Signal Detection in Graphs
A. Dynamic Integration
B. Residual Analysis
C. Dimensionality Reduction
IV. Applying the Framework Fundamentals:
A. Community characterization in social networks
B. Anomaly detection in information networks
C. Predicting interactions in dynamic networks

Biographies

Nadya T. Bliss is the Director of Strategic Projects Development at Arizona State University. In that role, she reports to the Senior Vice President, Knowledge Enterprise Development and is responsible for developing and maintaining strategic relationships with funding agencies, working with the University leaders on developing large--‐scale, interdisciplinary proposals, and monitoring University’s strategic investments. She holds a Professor of Practice Appointment in the College of Technology and Innovation and leads and participates in sponsored research. Prior to joining ASU in 2012, Nadya spent 10 years at MIT Lincoln Laboratory, most recently as the founding Group Leader of the Computing and Analytics Group. In that capacity, she developed, led, and managed research initiatives in advanced analytics, high performance computing systems, and computer architectures. Nadya was awarded the inaugural MIT Lincoln Laboratory Early Career Technical Achievement Award (2011) recognizing her work in parallel computing, computer architectures, and graph processing algorithms and her leadership in anomaly detection in graph--‐based data (presented to 2 employees under 35). She has over 65 publications and presentations, holds a patent, and is a Senior Member of IEEE. Nadya received bachelor (2002) and master degrees (2002) in Computer Science from Cornell University.

Alfred O. Hero III received the B.S. (summa cum laude) from Boston University (1980) and the Ph.D from Princeton University (1984), both in Electrical Engineering. Since 1984 he has been with the University of Michigan, Ann Arbor, where he is the R. Jamison and Betty Williams Professor of Engineering, in the Department of Electrical Engineering and Computer Science. In 2008 he was awarded the Digiteo Chaire d'Excellence, sponsored by Digiteo Research Park in Paris. He is a Fellow of the Institute of Electrical and Electronics Engineers (IEEE) and several of his research articles have received best paper awards. Alfred Hero was awarded the University of Michigan Distinguished Faculty Achievement Award (2011). He received the IEEE Signal Processing Society Meritorious Service Award (1998) and the IEEE Third Millennium Medal (2000). Alfred Hero was President of the IEEE Signal Processing Society (2006--‐2008) and was on the Board of Directors of the IEEE (2009--‐2011) where he served as Director of Division IX (Signals and Applications). Alfred Hero's recent research interests are in statistical signal processing, machine learning and the analysis of high dimensional spatio--‐temporal data with applications to networks, multi--‐modal sensing and tracking, database indexing and retrieval, imaging, and genomic signal processing.

Benjamin A. Miller received the B.S. degree (with highest honors) and the M.S. degree in computer science in 2005 from the University of Illinois at Urbana--‐Champaign. In 2005, he joined Lincoln Laboratory at the Massachusetts Institute of Technology as an Associate Staff member in the Embedded Digital Systems (later Embedded and High Performance Computing) group. In this role, he developed novel algorithms for real-time linearization of radio‐frequency electronics, researched methods and models for signal recovery in multi--‐sensor compressed sensing, and developed efficient spectral techniques for the detection of anomalies in large graphs. Since 2012, he has been a Technical Staff member of the Computing and Analytics group at Lincoln Laboratory, where his research continues to focus on the theoretical and computational aspects of anomaly detection in large networks and other dynamic, combinatorial structures. Mr. Miller is a member of the Institute of Electrical and Electronics Engineers (including the IEEE Signal Processing Society), the Association for Computing Machinery, and the Society for Industrial and Applied Mathematics. He holds 3 patents and is author or co--‐author of 30 peer--‐reviewed conference and journal papers on nonlinear signal processing, compressive sensing, and detection and estimation theory for graph--‐based data.


T2 - Monotone Operator Splitting Methods in Signal and Image Recovery
Subject Area: Image Processing
Speakers: P.L. Combettes (Université Pierre et Marie Curie – Paris 6), J.-C. Pesquet (Université Paris-Est), and N. Pustelnik (ENS de Lyon)

Summary
In recent years convex optimization has become the main thrust behind significant advances in signal processing, image processing and machine learning. The increasingly complex variational formulations encountered in these areas which may involve a sum of several, possibly non-smooth, convex terms, together with the large sizes of the problems at hand make the use of standard optimization methods such as those based on subgradient descent techniques intractable computationally. Since their introduction in the signal processing arena, splitting techniques have emerged as a central tool to circumvent these roadblocks: they operate by breaking down the problem into individual components that can be activated individually in the solution algorithm. By way of example, the success of these splitting techniques plays a prominent role for the solution of inverse problems involving sparse data (with applications in biomedical/satellite/seismic imaging), decompositions into meaningful components (e.g., noise-driven and low rank matrix components), or filter/pulse design. In the past decade, numerous convex optimization algorithms based on splitting techniques have been proposed or rediscovered in an attempt to efficiently deal with such problems. As shown by the recent publication of a book [1] on this topic, monotone operator theory provide a powerful framework to not only capture many of these algorithms in a simple framework (e.g., forward-backward, Douglas- Rachford, ADMM, SDMM splitting techniques) but also develop new efficient methods that could not have been devised by purely optimization-based arguments (e.g., recent primal-dual schemes using product space formulations involving skew operators). Monotone operator theory, a branch of nonlinear analysis that emerged in the early 1960s, makes it possible to reduce complex convex optimization problems to simple structured inclusion problems. It is worth noting that convex feasibility problems, which have been popular for a long time in signal and image processing, can be viewed as particular cases in this optimization framework. This tutorial aims at providing a comprehensive view of monotones operators and to show how they naturally enable the easy design of optimization algorithms based on the key concept of proximal operator originally introduced by Moreau. Three main algorithms (Forward-Backward, Douglas-Rachford and Forward-Backward-Forward) will be shown to be the basic building blocks of many other optimization methods. In particular, it will be emphasized that parallel strategies can be proposed which are efficiently implementable on multicore architectures. The presentation of the theoretical and algorithmic challenges will be illustrated with many examples from signal and image recovery.
[1] H. H. Bauschke and P. L. Combettes, Convex Analysis and Monotone Operator Theory in Hilbert Spaces, Springer, New York, 2011.

Outline
1-Background on monotones operators and interplay with convex analysis.
2-Nonexpansiveness, resolvent, and proximity operator.
3-Construction of a zero of a maximally monotone operator : Forward-backward, Douglas-Rachford splitting, and Forward-Backward-Forward splitting.
4-Main classes of proximal algorithms
5-Monotone operator, duality, and primal-dual proximal algorithms. Illustrations with inverse problem applications.

Biographies

Patrick L. Combettes (IEEE Fellow), received the PhD degree from North Carolina State University in 1989. In 1990, he joined the faculty of the City University of New York, where he became a Full Professor in 1999. Since September 1999, he has been with the faculty of mathematics of Université Pierre et Marie Curie, where he is presently a Full Professor (classe exceptionnelle). P.L. Combettes is a former associate editor of the IEEE Transactions on Image Processing, and is currently an associate editor of the SIAM Journal on Optimization, the Journal of Approximation Theory, and the Journal of Mathematical Analysis and Applications. He has organized several summer schools and workshops on the topics of optimization theory and mathematical image processing in France and abroad, and has published extensively on these topics. His main contributions to signal processing are the development of convexity methods and the introduction of proximal techniques.

Jean-Christophe Pesquet (IEEE Fellow) received the engineering degree from Supélec, Gif-sur-Yvette, France (1987), the Ph.D. and HDR degrees from the University Paris-Sud, Paris, France (1990 and 1999), respectively. From 1991 to 1999, he was an Assistant Professor at the University Paris-Sud. He is currently a Full Professor (classe exceptionnelle) with University Paris-Est Marne-la-Vallée, France and the Deputy Director of the Laboratoire d'Informatique Gaspard Monge. In 2005, J.-C. Pesquet was technical chairman of the IEEE ICASSP conference. He served as an associate editor for the IEEE Signal Processing Letters journal. He was an associate editor for the IEEE Transactions on Signal Processing and he is currently a senior area editor for the same journal and a member of the committee for the best paper award of the EURASIP Journal on Advances in Signal Processing. He is also currently a member of the Signal Processing Theory and Methods technical committee of the IEEE Signal Processing Society. Prof. Pesquet published numerous papers in international journals and conferences. His research interests include multiscale analysis, wavelets, filter banks, statistical signal processing, inverse problems and optimization with applications to imaging. His work received 4 awards from EURASIP and IEEE.

Nelly Pustelnik received the engineering degree from Institut des Sciences de l'Ingénieur de Toulon et du Var and a Master degree from Université de Toulon in 2007. She received the PhD degree in signal and image processing from the University Paris-Est Marne-la-vallée, France, in 2010. In 2011, she was with the Laboratoire IMS, Université de Bordeaux, France, where she held a post-doctoral position work - ing on convex optimization and limited view angle tomograhy as part of a collaboration with TOTAL SA. Since 2012, she is a CNRS researcher in the Signal Processing Team of the Laboratoire de Physique de l'ENS de Lyon. Her activity is focused on non-smooth convex optimization and its applications in signal and image processing (e.g. mode decomposition, image restoration, texture segmentation). Dr. Pustelnik is a reviewer for IEEE Trans. Image Proc., IEEE Trans. Signal Proc, SIAM Journal on Imaging Sciences, Inverse problems, IEEE ICASSP 2013, SPARS'2013 and she was involved in the Technical Program committee of EUSIPCO 2012 and 2013. She is a part of the organizing committee of IEEE ICIP 2014.


T3 - Informed Audio Source Separation: Trends, Approaches and Algorithms
Subject Area: Speech/Audio/Language Processing
Speakers: Alexey Ozerov (Technicolor), Antoine Liutkus (INRIA, Nancy Grand Est) and Gaël Richard (Telecom ParisTech)

Summary
Source separation, and more precisely audio source separation as we consider in this tutorial, remains today challenging in many cases, especially in the undetermined case, when there are less observations than sources, and in the single-channel case. The main difficulty is usually due to the lack of information about latent sources. Thus, in order to boost audio source separation performance, many recent works (mostly in the period of the past 3-4 years) turned towards so-called informed audio source separation approaches, where the separation algorithm relies on any kind of additional information about sources (e.g., corresponding musical score or text) to better extracting them. Moreover, several trends that can be differentiated by the way this additional information is produced were recently developed in parallel. The goal of this tutorial is to make a comprehensive review of three major trends in informed audio source separation and to present in details the corresponding methods and algorithms. After a brief introduction in audio source separation we discuss in details the following three trends in informed audio source separation: Auxiliary data-informed source separation: The additional information is assumed to be available which is a reasonable assumption in many cases. For example, this can be a musical score corresponding to the musical source to be separated, or a video of a speaking person corresponding to the speech signal to be extracted. User-guided source separation: The additional information is created by a user itself with the intention to improve the source separation. For example, this can be some indication about source activity in the time-frequency domain. In contrast to the previous trend, where the information is just given and unchanged, the user can possibly correct remaining separation errors, e.g., by listening to both the mixture and the separation result and by modifying the provided information accordingly. Coding-based informed source separation: The additional information is created by an algorithm at a so-called encoding stage where both the sources and the mixtures are assumed known. This information is required to be small and should help the source separation at a so-called decoding stage, where the sources are no longer assumed to be known. This trend is at the crossroads of source separation and compression, and it shares many similarities with the recently introduced Spatial Audio Object Coding (SAOC). We conclude by discussing the remaining challenges and by drawing some further research directions.

Biographies

Alexey Ozerov holds a Ph.D. in Signal Processing from the University of Rennes 1 (France). He worked towards this degree from 2003 to 2006 in the labs of France Telecom R&D and in collaboration with the IRISA institute. Earlier, he received an M.Sc. degree in Mathematics from the Saint-Petersburg State University (Russia) in 1999 and an M.Sc. degree in Applied Mathematics from the University of Bordeaux 1 (France) in 2003. From 1999 to 2002, Alexey worked at Terayon Communicational Systems (USA) as a R&D software engineer, first in Saint-Petersburg and then in Prague (Czech Republic). He was for one year (2007) in Sound and Image Processing Lab at KTH (Royal Institute of Technology), Stockholm, Sweden, for one year and half (2008-2009) in Télécom ParisTech / CNRS LTCI - Signal and Image Processing (TSI) Department, and for two years (2009 - 2011) with METISS team of IRISA / INRIA - Rennes. Now he is with Technicolor Research & Innovation at Rennes, France. His research interests include audio source separation, source coding, automatic speech recognition and recently image processing.

Antoine Liutkus was born in France on February 23rd, 1981. He received the State Engineering degree from Telecom ParisTech, France, in 2005, along with the M.Sc. degree in acoustics, computer science and signal processing applied to music from the Université Pierre et Marie Curie (Paris VI). He worked as a research engineer on source separation at Audionamix from 2007 to 2010 and obtained his Ph.D. degree in the Department of Signal and Image Processing, Télécom ParisTech in 2012. Today, he is a research associate at INRIA Nancy Grand Est, France. His research interests include statistical signal processing, source separation, inverse problems and machine learning methods applied to signal processing.

Gaël Richard received the State Engineering degree from Télécom ParisTech, in 1990 and the Ph.D. degree from University of Paris-XI, in 1994 in speech synthesis. He then spent two years at the CAIP Center, Rutgers University, in the Speech Processing Group of Prof. J. Flanagan, where he explored innovative approaches for speech production. From 1997 to 2001, he successively worked for Matra-Nortel, Philips and L&H. In particular, he was the Project Manager of several large scale European projects in the field of speech, audio and multimodal signal processing. In 2001, he joined Télécom ParisTech, where he is now a Full Professor in audio signal processing and Head of the Audio, Acoustics, and Waves research group. He is a coauthor of over 200 papers and inventor in a number of patents. He was an Associate Editor of the IEEE Transactions on Audio, Speech and Language Processing between 1997 and 2011 and one of the guest editors of the special issue on ``Music Signal Processing'' of IEEE Journal on Selected Topics in Signal Processing (2011). He currently is a member of the IEEE Audio and Acoustic Signal Processing Technical Committee, member of the EURASIP and AES and senior member of the IEEE.


T4 - Signal Processing for Analog Systems
Subject Area: Signal Processing System Design and Implementation
Speakers: Arthur J. Redfern, Manar El-Chammas and Lei Ding (Texas Instruments)>

Summary
It’s hard to overstate the importance of the Nyquist-Shannon sampling theorem to the field of signal processing. The ability to represent an analog signal with a finite number of discrete samples has successfully allowed researchers in many areas to focus on the design of discrete algorithms that run on digital processors. Increasingly, however, it’s the devices at the analog and digital interfaces that dominate the power budget and performance of the overall system.

In this tutorial we turn our attention to these devices and show the past accomplishments and future opportunities for applying signal processing directly to the design of analog devices that sit at the boundary of the analog and digital worlds. Specific items this tutorial will accomplish include:
- Provide a view of the analog design process specifically developed for people with a signal processing background to elucidate what is practically possible, describe fundamental tradeoffs and point out trends.
- Provide a conceptual view of what can be accomplished with analog design alone, with signal processing tacked onto an existing analog design and with signal processing and analog design tools used together from the beginning to develop an optimized system design.
- Show specific examples and key accomplishments where joint signal processing and analog design have been used to realize systems in the areas of ADCs, DACs and amplifiers.
- Give insight into places where there is room for improvement and places where there’s less to be gained because existing designs are already pushing against fundamental limits.
Along the way we’ll show how familiar concepts from signal processing – information theory, channel capacity, blind identification, compressive sensing, nonlinear signal processing and big data – all fit with and apply to the design of analog devices.

Biographies

Arthur J. Redfern received a B.S. in 1995 from the University of Virginia and a M.S. and Ph.D. in 1996 and 1999, respectively, from the Georgia Institute of Technology, all in electrical engineering. Following his thesis work on nonlinear systems modeled by the Volterra series, Arthur joined the Systems and Applications R&D Center at Texas Instruments where he currently manages the Signal Processing for Analog Systems branch. His activities at TI have spanned the areas of analog (ADCs, AFEs, amplifiers, antennas, DACs and design optimization), communications (DSL, DTV and SerDes) and signal processing (speakers and touch screens). He has over 20 papers published in refereed conferences and journals and has been granted over 20 US patents.

Manar El-Chammas received the B.E. degree in Computer and Communications Engineering from the American University of Beirut, Lebanon, in 2004, and the M.S. and Ph.D. degrees in Electrical Engineering from Stanford University, Stanford, CA, in 2006 and 2010, respectively. He has held internship positions with the Advanced Development Lab, LSI Corporation, the IBM Zurich Research Lab, and Kilby Labs in Texas Instruments. He currently works at Texas Instruments, Inc., Dallas, TX on high-speed data converters. His research interests include mixed-signal integrated circuit design, with emphasis on data converters and background calibration.

Lei Ding received the B.S. degree in Electrical Engineering from the Nanjing University of Aeronautics and Astronautics, Nanjing, China, in 1997, the M.S. degree in Biomedical Engineering from the University of Alabama at Birmingham in 2000. In 2004, he received the Ph.D. degree in Electrical Engineering from the Georgia Institute of Technology, Atlanta. From 2004 to 2008, he was a Digital Design Engineer with Cirrus Logic, Austin, TX, designing low-power audio data converters. Since 2008, he has been with the Embedded Processing Systems Lab, Texas Instruments, Dallas, TX, and is currently a Systems Engineer working in the area of signal processing design and implementation for analog systems. He holds 26 US patents (issued and pending) and has published over 20 papers.


T5 - Transmitter Cooperation in Wireless Networks: Potential and Challenges
Subject Area: Communications
Speakers: David Gesbert and Paul de Kerret (EURECOM)

Summary
The traffic of wireless data, which has steadily increased in the past decade, is poised to explode in view of the additional internet and multimedia services to be deployed by operators. To face with this demand, system designers are adopting an array of strategies, the leading of which simply consists in bringing the infrastructure closer to the end user by densifying the network and making it adaptive to local traffic conditions (macro-,micro-, pico-, femto-cells etc.). As densification combined with an aggressive spectrum reuse policy brings increasing amounts of multi-cell interference, approaches to mitigate interference relying on cooperation or coordination among wireless nodes have gained extraordinary levels of attention in both the standardization and research communities. Although cooperation and coordination can be carried over a variety of communications parameters (time/frequency scheduling, power control, antenna combining, etc.) a common feature between all existing approaches lies in the need for cooperating nodes to learn about the environment (channel state) and exchange this information among themselves.

The acquisition and exchange of timely channel state information represents in many cases the single most important hurdle threatening the scalability of coordination and cooperation methods and ultimately their practical adoption. As a countermeasure, the concept of distributed coordination or cooperation, by which nodes try to coordinate on the basis of mostly local information or via the exchange of limited information, has now emerged as a rich topic of research in its own right.

In this tutorial, we start by presenting the fundamental signal processing approaches for managing interference in wireless networks. Then, we discuss the state of the art methods enabling cooperation in practical networks and their limitations. In the second part of the tutorial we turn our attention on the general framework of coordination with distributed channel state information. Complying with a basic assumption in practical networks, we assume that the communications between nodes that seek coordination is restricted. We demonstrate how this problem can be recast into a general so-called “Team Decision formulation”. We show examples of solutions to the problem of distributed precoding for application to the CoMP/Network MIMO framework. The problem of optimally allocating spatially the channel state information is also studied. Finally, interesting open problems are presented.

Outline
I. Fundamentals for Interference Management
A. Single transmitter schemes
B. Transmitter cooperation (Network MIMO, coordinated beamforming, interference alignment, coordinated resource allocation)
C. Trends in standardization
II. Transmitter Cooperation in Practical Networks: Challenges, Conventional Solutions and Limitations
A. Clustering
B. Hardware impairments
C. Channel estimation
D. Limited feedback
E. Exchange of channel state information
III. Coordination with Distributed Channel State Information (CSI)
A. The team decision perspective
B. Coordination with limited information exchange
IV. Precoding design with Imperfect and Imperfectly Shared CSI
A. Impact of distributed CSIT
B. Precoder design with distributed CSI
V. Spatial allocation of feedback resources
A. Who needs to know what?
B. Interference alignment with incomplete CSIT sharing
C. Network MIMO with distance based CSIT allocation
VI. Open Problems

Biographies

David Gesbert (IEEE Fellow) is Professor in the Mobile Communications Dept., EURECOM, where he heads the Mobile Communication Dept. He obtained the Ph.D degree from Ecole Nationale Superieure des Telecommunications, France, in 1997. From 1997 to 1999 he has been with the Information Systems Laboratory, Stanford University. In 1999, he was a founding engineer of Iospan Wireless Inc, San Jose, Ca., a startup company pioneering MIMO-OFDM (now Intel). Between 2001 and 2003 he has been with the Department of Informatics, University of Oslo as an adjunct professor. D. Gesbert has published about 200 papers and several patents all in the area of signal processing, communications, and wireless networks. D. Gesbert was a co-editor of several special issues on wireless networks and communications theory, for JSAC (2003, 2007, 2009), EURASIP Journal on Applied Signal Processing (2004, 2007), Wireless Communications Magazine (2006). He served on the IEEE Signal Processing for Communications Technical Committee, 2003-2008. He is an associate editor for IEEE Transactions on Wireless Communications and the EURASIP Journal on Wireless Communications and Networking. He authored or co-authored papers winning the 2013 Best Signal Processing Magazine Paper, the 2004 IEEE Best Tutorial Paper Award (Communications Society) for a 2003 JSAC paper on MIMO systems, 2005 Best Paper (Young Author) Award for Signal Proc. Society journals, and the Best Paper Award for the 2004 ACM MSWiM workshop. He co-authored the book “Space time wireless communications: From parameter estimation to MIMO systems”, Cambridge Press, 2006.

Paul de Kerret (IEEE Student Member) graduated in 2009 from Ecole Nationale Superieure des Telecommunications de Bretagne, France and obtained a diploma degree in electrical engineering from Munich University of Technology (TUM), Germany. He also earned a four year degree in mathematics at the Universite de Bretagne Occidentale, France in 2008. From January 2010 to september 2010, he has been a research assistant at the Institute for Theoretical Information Technology, RWTH Aachen University, Germany. He is currently completing his Ph.D. degree in the Mobile Communications Department at EURECOM, France. During his doctorate, he has been involved in two European projects with both academic and industrial partners where the challenges of the future wireless networks were discussed in terms of academic research and in terms of standardization. He has also worked in cooperation with several elite research centers and universities in Europe. He is the author of journal papers submitted to prestigious journals and a magazine focused on the cooperation of transmitters in wireless networks.


T6 - Signal Processing for Big Data
Subject Area: Fundamentals
Speakers: G.B. Giannakis, Konstantinos Slavakis (University of Minnesota), Gonzalo Mateos (Carnegie Mellon University)

Summary
We live in an era of data deluge. Pervasive sensors collect massive amounts of information on every bit of our lives, churning out enormous streams of raw data in various formats. Mining information from unprecedented volumes of data promises to limit the spread of epidemics and diseases, identify trends in financial markets, learn the dynamics of emergent social-computational systems, and also protect critical infrastructure including the smart grid and the Internet’s backbone network. While Big Data can be definitely perceived as a big blessing, big challenges also arise with large-scale datasets. The sheer volume of data makes it often impossible to run analytics using a central processor and storage, and distributed processing with parallelized multi-processors is preferred while the data themselves are stored in the cloud. As many sources continuously generate data in real time, analytics must often be performed “on-the-fly” and without an opportunity to revisit past entries. Due to their disparate origins, massive datasets are noisy, incomplete, prone to outliers, and vulnerable to cyber-attacks. These effects are amplified if the acquisition and transportation cost per datum is driven to a minimum. Overall, Big Data present challenges in which resources such as time, space, and energy, are intertwined in complex ways with data resources. Given these challenges, ample signal processing opportunities arise. This tutorial seeks to provide an overview of ongoing research in novel models applicable to a wide range of Big Data analytics problems, as well as algorithms and architectures to handle the practical challenges, while revealing fundamental limits and insights on the mathematical trade-offs involved.    

Biographies

G. B.  Giannakis (Fellow'97) received his Diploma in Electrical Engr. from the Ntl. Tech. Univ. of Athens, Greece, 1981. From 1982 to 1986 he was with the Univ. of Southern California (USC), where he received his MSc. in Electrical Engineering, 1983, MSc. in Mathematics, 1986, and Ph.D. in Electrical Engr., 1986. Since 1999 he has been a professor with the Univ. of Minnesota, where he now holds an ADC Chair in Wireless Telecommunications in the ECE Department, and serves as director of the Digital Technology Center. His general interests span the areas of communications, networking and statistical signal processing - subjects on which he has published more than 360 journal papers, 590 conference papers, 20 book chapters, two edited books and two research monographs (h-index 105). Current research focuses on sparsity and big data analytics, wireless cognitive radios, mobile ad hoc networks, renewable energy, power grid, gene-regulatory, and social networks. He is the (co-) inventor of 21 patents issued, and the (co-) recipient of 8 best paper awards from the IEEE Signal Processing (SP) and Communications Societies, including the G. Marconi Prize Paper Award in Wireless Communications. He also received Technical Achievement Awards from the SP Society (2000), from EURASIP (2005), a Young Faculty Teaching Award, and the G. W. Taylor Award for Distinguished Research from the University of Minnesota. He is a Fellow of EURASIP, and has served the IEEE in a number of posts, including that of a Distinguished Lecturer for the IEEE-SP Society.

Konstantinos Slavakis (SM’12) was born in Thessaloniki, Greece. He received the M.Eng. and Ph.D. degrees in Electrical and Electronic Engineering from Tokyo Institute of Technology (TokyoTech), Japan, in 1999 and 2002, respectively. He has been a Japanese Government Scholar (1996-2002), a JSPS Postdoctoral Fellow at TokyoTech (2004-2006), and a PostDoctoral Fellow with the Dept. of Informatics and Telecommunications, University of Athens, Greece (2006-2007). He has served as an Assistant Professor with the Department of Telecommunications and Informatics, at the University of Peloponnese, Greece (2007-2012), and he is currently a Research Associate Professor with the Dept. of ECE, and the Digital Technology Center at the University of Minnesota, USA. His research interests span applications of convex analysis and computational algebraic geometry to signal processing, machine learning, big data analytics, and multidimensional systems’ problems. He has been serving IEEE Trans. on Signal Processing as Associate Editor (2009-2013), and as Senior Area Editor since 2010.

Gonzalo Mateos was born in Montevideo, Uruguay, in 1982. He received his B.Sc. degree in Electrical Engineering from Universidad de la Republica, Uruguay, in 2005, and the M.Sc. and Ph.D. degrees in Electrical Engineering from the University of Minnesota (UofM), Twin Cities, in 2009 and 2011. From 2004 to 2006, he worked as a Systems Engineer at Asea Brown Boveri (ABB), Uruguay. Currently, he is a visiting researcher with the Computer Science Department at Carnegie Mellon University. His research interests lie in the areas of statistical learning from Big Data, network science, wireless communications, and signal processing. His current research focuses on algorithms, analysis, and application of statistical signal processing tools to dynamic network health monitoring, social, power grid, and Big Data analytics. Since 2012, he serves on the Editorial Board of the EURASIP Journal on Advances in Signal Processing. He received the Best Student Paper Award at the 13th IEEE Workshop on Signal Processing Advances in Wireless Communications, 2012 held at Cesme, Turkey, and was also a finalist of the Student Paper Contest at the 14th IEEE DSP Workshop, 2011 held at Sedona, Arizona, USA. He received a honorable mention for the UofM's Best Dissertation Award program across all Physical Sciences and Engineering areas.


T7 - Semidefinite Relaxation: From Theory to Applications to Latest Advances
Subject Area: Fundamentals
Speakers: Wing-Kin Ma and Anthony Man-Cho So (The Chinese University of Hong Kong)

Summary
In recent years, semidefinite relaxation (SDR) has emerged as an important and versatile tool for tackling various problems in signal processing and communications. Originally employed as an efficient heuristic for approximating difficult optimization problems—such as maximum-likelihood MIMO detection, which is NP-hard—SDR has now been proven to have much broader impact than just being a computational tool. Indeed, recent developments in topics such as sensor network localization and transmit beamforming have demonstrated clearly that SDR can provide not only new design insights but also new, computationally efficient models for key frontier issues in signal processing.

Given the maturity of SDR and the role it plays in signal processing and communications, the goal of this tutorial is to give an overview of SDR–from the basic concepts to some of the more sophisticated theoretical results, and from the classical applications to the forefront advances. Our emphasis will be on extracting insights from theory, with an aim of interpreting their practical implications. To showcase the power of SDR, we will focus on two applications; namely, transmit beamforming and sensor network localization. In particular, transmit beamforming has attracted much interest lately and has led to a number of exciting advances even to SDR theory itself. Some new fronts in the field–such as semidefinite programming (SDP) rank reduction theory; a rank-two SDR generalization of the presently popularized rank-one SDR beamforming framework; chance-constrained optimization and its application to outage-based robust transmit beamforming; polynomial-time methods for SDPs with non-convex regularizations and their application to rank-constrained SDPs–will also be covered in this tutorial.

Outline
1)SDR Technique
a)Basic concepts, practical deployment
b)Provable approximation accuracies
c)Application example: MIMO detection
d)SDP rank reduction, and the optimality of SDR

2)Theme Application I: Transmit Beamforming
a)Rank-one SDR beamforming framework
b)Basic problems: unicast and multicast beamforming
c)SDP rank reduction and tightness of SDR
d)Extensions:
d.i)discrete sum rate maximization
d.ii)multicell-coordinated beamforming, decentralized optimization
d.iii)spectrum sharing in cognitive radios
d.iv)one-way and two-way relaying
d.v)physical-layer secrecye
d.vi)energy harvesting
e)Advanced:
e.i)rank-two SDR and beamformed Alamouti space-time coding, provable approximation accuracies
e.ii)outage-constrained robust beamforming via chance-constrained optimization

3)Theme Application II: Sensor Network Localization
a)Basic problem: range-based localization
b)Fast heuristics
c)Extension: localization based on other measurements, noisy measurements, and/or imperfect anchor locations
d)Advanced: polynomial-time methods for SDPs with non-convex penalties


Biographies

Wing-Kin Ma received the B.Eng. degree in electrical and electronic engineering from the University of Portsmouth, Portsmouth, U.K., in 1995 and the M.Phil. and Ph.D. degrees, both in electronic engineering, from the Chinese University of Hong Kong (CUHK), Hong Kong, in 1997 and 2001, respectively. He is currently an Associate Professor with the Department of Electronic Engineering, CUHK. His research interests are in signal processing and communications, with a recent emphasis on MIMO, optimization, and blind signal processing. Dr. Ma is currently serving or has served as Associate Editor and Guest Editor of several journals. He is an SPTM-TC Member.

Anthony Man-Cho So received his BSE degree in Computer Science from Princeton University in 2000. He then received his MSc and PhD degree in Computer Science from Stanford University in 2002 and 2007, respectively. He joined The Chinese University of Hong Kong (CUHK) in 2007, where he currently serves as Assistant Dean of the Faculty of Engineering and is Associate Professor in the Department of Systems Engineering and Engineering Management. Professor So is a recipient of the 2010 Optimization Prize for Young Researchers awarded by the Optimization Society of the Institute for Operations Research and the Management Sciences (INFORMS).


T8 - EEG Signal Processing and Classification for Brain Computer Interfacing (BCI) Applications
Subject Area: Biomedical signal processing
Speakers: Amit Konar (Jadavpur University), Fabien Lotte (INRIA-Bordeaux Sud-Ouest), Arijit Sinharay (Tata Consultancy Services Ltd)

Summary
Brain-Computer Interfacing (BCI) provides an intelligent channel of communication between the human brain and a computer with an ultimate aim to control brain-states using artificially generated stimuli and/or to decode brain states involving attention, perception, motor imagination or any other cognitive functioning. ElectroEncephaloGraphy (EEG) is particularly interesting for picking up brain signals because of its non-invasive nature and portability. This tutorial aims at presenting the current state-of-the art research direction in EEG signal processing and classification techniques for various BCI applications. It begins with a review of the functionalities of different areas of the human brain to demonstrate their possible scope of application in EEG localization. The deliberation then considers the acquisition of EEG signals with a discussion of the choice of EEG modalities, such as, P300, ERD/ERS, SSVEP, ErrP etc. for use in different applications. Then the tutorial will discuss in more details EEG signal pre-processing and filtering, in order to enhance the relevant information embedded in EEG signals, and filter out taskunrelated signals. More particularly, this part will focus on spectral and spatial filtering of EEG signals such as inverse solution, Common Spatial Patterns or xDAWN algorithms. Next, the presentation focuses on different feature selection techniques including PCA, ICA, evolutionary feature selection as performance of EEG-BCI system greatly depends on feature selection and classification. Usually, supervised classifiers are used in EEG-BCI applications. The list of classifiers to be discussed includes Naïve Bayes, SVM, Expectation Maximization, ADABOOST and a few other with their relative merits and demerits. Lastly, we introduce the scope of applications of the proposed methodologies in several interesting applications. The domain problems considered includes communication and control for paralyzed users, cognitive load detection, olfactory stimulus classification, artificial Limb control, true emotion vs. pretension detection, BCI-based control of virtual reality and gaming, stress detection etc. Finally, the presentation comes to an end with a brief discussion on future direction of EEG-BCI signal processing and related applications.

Amit Konar received his Ph.D. (Engineering) degree in Artificial Intelligence from Jadavpur University (JU), Kolkata, India in 1994. He is currently a Professor in the department of Electronics and Tele- Communication Engineering, JU and Joint Coordinator, Center for Cognitive Science, JU. He is also the founding coordinator of a regular M.Tech. program on Intelligent Automation and Robotics, offered by JU. He has published over 300 papers and eight books on different sub-areas of machine intelligence. Dr. KONAR’s current research includes physiological signal processing to recognize the psychological states of the human mind. He is currently the Principal Investigator of two major nationally acknowledged research projects on decoding of motor imagery for artificial limb control, and Olfactory perceptualability measurement by EEG-analysis. He is also a co-author of a forthcoming book on Non-Invasive Human-Computer Interface for Rehabilitative Applications, Bio-systems and Bio-robotics Series, Springer. Dr. KONAR serves as an Associate Editor of IEEE Trans. on Fuzzy Systems, Neurocomputing, Elsevier, and Int. J. of Intelligent Decision Technologies, IOS Press. He also served as the Editor-in-Chief of Int. J. of Artificial Intelligence and Soft Computing, Inderscience, and as Associate Editor of IEEE Trans. on Systems, Man and Cybernetics: Part-A for three years till December 2012.

Fabien Lotte obtained a M.Sc. (2005) and a PhD degree (2008) in computer sciences, from INSA Rennes, France. His PhD Thesis work on Brain-Computer Interfaces (BCI) received both the PhD Thesis award 2009 from AFRIF (French Association for Pattern Recognition) and the PhD Thesis award 2009 accessit (2nd prize) from ASTI (French Association for Information Sciences and Technologies). In 2009 and 2010, he was a research fellow in the BCI Laboratory at the Institute for Infocomm Research in Singapore. There, he explored EEG signal processing for the design of robust BCI. Since January 2011, he is a Research Scientist (with tenure) at Inria Bordeaux Sud-Ouest, in France, where he works on BCI, user interaction and signal processing. He has authored more than 40 papers on BCI, including a dozen publications in some of the best journals in the field (IEEE Transactions on Biomedical Engineering, IEEE Transactions on Signal Processing, Journal of Neural Engineering, etc.). He also gave several tutorials on BCI in international conferences and schools (APSIPA ASC 2010, Neurocomp 2011, BBCI Summer school on Neurotechnology 2012). Finally, he is a review editor for Frontiers in Neuroprosthetics, a first-tier electronic open access journal dedicated to BCI research.

Arijit Sinharay graduated from Portland State University (PSU), US with MS in Physics and MS in Electrical & Computer Engineering in 2003. He received the Patty Jeane Semura Outstanding Graduate Student Award from the Physics Department in 2002. Immediately after his graduation he joined The Adept Group, Inc, Los Angeles, US as an R&D Electronics Engineer where he had extensively worked on ultrasonic sensors and related signal processing areas. On 2005, he joined Innovation labs of Tata Consultancy Services, Kolkata, India as a Research Engineer. Currently he leads the BCI research in the Innovation Lab, Kolkata. In his 10 years of work experience he had worked on diverse fields including wireless communication, bio-medical signal processing, industrial sensors, Bio-sensors, and Scanning Microscopy with primary focus on signal processing, machine learning and non-invasive sensor developments.


T9 - Deep learning for natural language processing and related applications
Subject Area: Speech/Audio/Language Processing
Speakers: Xiaodong He, Jianfeng Gao, Li Deng (Microsoft Research)

Summary
Deep learning techniques have enjoyed tremendous success in the speech and language processing community in recent years (especially since 2011), establishing new state-of-the-art performance in speech recognition, language modeling, and some natural language processing tasks. The focus of this tutorial is on deep learning approaches to problems in language or text processing, with particular emphasis on important applications including spoken language understanding (SLU), machine translation (MT), and semantic information retrieval (IR) from text.

In this tutorial, we first survey the latest deep learning technology, presenting both theoretical and practical perspectives that are most relevant to our topic. We plan to cover common methods of deep neural networks and more advanced methods of recurrent, recursive, stacking, and convolutional networks. Next, we review general problems and tasks in text/language processing, and underline the distinct properties that differentiate language processing from other tasks such as speech and image object recognition. More importantly, we highlight the general issues of language processing, and elaborate on how new deep learning technologies are proposed and fundamentally address these issues. We then place particular emphasis on three important applications:1) spoken language understanding, 2) machine translation, and 3) semantic information retrieval from text. For each of the three tasks we discuss what particular architectures of deep learning models are suitable given the nature of the task, and how learning can be performed efficiently and effectively using end-to-end optimization strategies.

Beyond providing a systematic tutorial of the general theory, we also present hands-on experience in building state-of-the-art SLU/MT/IR systems. In the tutorial, we will share our practice with concrete examples drawn from our first-hand experience in major research benchmarks and some industrial scale applications which we have been working on extensively in recent years.

Outline
1.Background of deep learning and commonly used models
a.i.Background: A review
a.i.1.Review deep learning theory and progress in NLP & other relevant areas including speech and image processing
a.ii.Deep neural networks for modeling abstract meaning of text
a.ii.1.Deep auto-encoders
a.ii.2.Deep stacking neural networks
a.iii.Advanced architectures for modeling language structure
a.iii.1.Recurrent neural networks
a.iii.2.Recursive neural networks
a.iii.3.Temporal convolutional neural networks

2.General deep learning techniques for language processing
a.i.Common problems in language processing: Why deep learning is needed?
a.ii.Semantic representation
a.ii.1.Word embedding
a.ii.2.Semantic hashing
a.ii.3.Applications to language modeling
a.iii.Learning techniques

3.Deep learning in spoken language understanding
a.i.Overview of SLU
a.ii.Semantic classification using DCN and kernel-DCN
a.iii.Slot filling using RNN, bi-directional RNN, and embedding

4.Deep learning in machine translation
a.i.Overview of MT
a.ii.Translation model: NN based semantic translation model
a.iii.Language model: RNN based LM and its application to MT

5.Deep learning in information retrieval
a.i.Overview of IR
a.ii.The use of word hashing
a.iii.Deep structured semantic models (DSSM) for IR
a.iv.Convolutional DSSM for IR

6.Summary and discussion
a.i.Ongoing and future advances in deep learning for language processing


Biographies

Xiaodong He is a Researcher of Microsoft Research, Redmond, WA, USA. He is also an Affiliate Professor in Electrical Engineering at the University of Washington, Seattle, WA, USA. His research interests include speech recognition, spoken language understanding, machine translation, natural language processing, information retrieval, and machine learning. Dr. He has published a book and more than 60 technical papers in these areas, and has given a tutorial on speech translation at ICASSP2013. In benchmark evaluations, he and his colleagues have developed entries that obtained No. 1 place in the 2008 NIST Machine Translation Evaluation (NIST MT) and the 2011 International Workshop on Spoken Language Translation Evaluation (IWSLT), both in Chinese-English translation, respectively. He served as Associate Editor of IEEE Signal Processing Magazine, Guest Editor of IEEE Transactions on Audio, Speech and Language Processing, and Lead Guest Editor of IEEE Journal of Selected Topics in Signal Processing. He served in the organizing committee of ICASSP2013 as the Chair of Special Sessions, and general Co-Chair of the Workshop on Speech and Language at NIPS 2008. He was in program committees of major speech and language processing conferences. He is a senior member of IEEE and a member of ACL.

Jianfeng Gao is a Principal Researcher of Microsoft Research, Redmond, WA, USA. His research interests include Web search and mining, information retrieval, natural language processing and statistical machine learning. Dr. Gao has published more than 100 technical papers in these areas. He gave a tutorial on statistical translations and web search at 2011 ACL/SIGIR Summer School. In benchmark evaluations, he and his colleagues have developed entries that obtained No. 1 place in the 2008 NIST Machine Translation Evaluation (NIST MT) in Chinese-English translation. He was Associate Editor of ACM Trans on Asian Language Information Processing, (2007 to 2010), and was Member of the editorial board of Computational Linguistics (2006 – 2008). He also served as tutorial co-chair for CIKM2013, and area chairs for ACL2012, EMNLP 2010, ACL-IJCNLP 2009, etc.

Li Deng is a Principal Researcher of Microsoft Research Redmond. In the general areas of audio/speech/language technology and science, machine learning, signal/information processing, and computer science, he has published over 300 refereed papers in leading journals and conferences and 4 books. He is a Fellow of the Acoustical Society of America, a Fellow of the IEEE, and a Fellow of the International Speech Communication Association. He served on the Board of Governors of the IEEE Signal Processing Society (2008-2010). More recently, he served as Editor-in-Chief for the IEEE Signal Processing Magazine (2009-2011), which earned the highest impact factor in 2010 and 2011 among all IEEE publications and for which he received the 2012 IEEE SPS Meritorious Service Award. He recently served as General Chair of the IEEE ICASSP-2013, and currently serves as Editor-in-Chief for the IEEE Transactions on Audio, Speech and Language Processing. In 2009, in collaboration with Geoff Hinton, he initiated deep learning research at Microsoft. Since then, his technical work on and the leadership in industry-scale deep learning with colleagues and academic collaborators have created significant impact in speech recognition and in other areas of signal/information processing including text processing.


T10 - Bits and Flops in modern communications: analyzing complexity as the missing piece of the wireless-communication puzzle
Subject Area: Communications
Speakers: Petros Elia (EURECOM) and Joakim Jaldén (Royal Institute of Technology, KTH, Sweden)

Summary
The tutorial offers an exposition of newly established rigorous relationships between telecommunications-performance and computational-complexity, with emphasis on outage limited high rate MIMO scenarios. We provide exciting new results, and place strong emphasis on future challenges. We argue that complexity is the missing piece of the communication puzzle, and make the point that these challenges strike at the core of the theory and practice of modern wireless communications. The basic premise of the tutorial is that there is an urgent need to develop a unified theory, techniques and algorithms, for analyzing and minimizing the overall implementation complexity of future communication networks, and to decipher and practically achieve the fundamental performance-complexity limits achievable by collectives of communicating nodes. This tutorial comes at a time where recent advances in telecommunications systems have provided powerful communications algorithms that allow for large reductions in transmission power, at the expense though of exponential increases in computational complexity, and consequently in prohibitive increases in algorithmic power consumption. In conjunction with a broad introduction of recent work, the tutorial will reveal future challenges and open problems relating to:
* What is the complexity price to pay for near-optimal implementation of MIMO, multiuser and cooperative communications?
* Why is complexity a bottleneck?
* What are meaningful complexity measures?
* What policies can regulate complexity at a limited performance loss?
* How does complexity-constraints affect cooperative protocols, what is the best protocol, and how many relays does it involve?
* How should multiple users behave in the presence of complexity constraints?
* How big of a chip do you need to cancel interference?
* How does feedback reduce complexity?
* How can we convert the non-ergodic MIMO channel to an ergodic channel with one bit of feedback and a few flops?
The tutorial is designed to attract audience from many branches of communications, while the presented framework can be general enough to attract researchers in areas where performance comes at the expense of non-negligible computational costs. The tutorial will conclude with an open ended exposition of what is involved in broadening the results to communications scenarios other than the outage limited MIMO setting.

Biographies

Petros Elia received the M.Sc. and Ph.D. in electrical engineering from the University of Southern California (USC), Los Angeles, in 2001 and 2006 respectively. Since February 2008 he has been an Assistant Professor with the Department of Mobile Communications at EURECOM in Sophia Antipolis, France. His research interests include combining approaches from different sciences, such as mathematics, physics, and from information theory, complexity theory, and game theory, towards analysis and algorithmic design for distributed and decentralized communication networks. His latest research deals with MIMO, cooperative and multiple access protocols and transceivers, complexity of communication, isolation and connectivity in dense networks, queueing theory and cross-layer design, coding theory, information theoretic limits in cooperative communications, and surveillance networks.  He is a Fulbright scholar, the co-recipient of the SPAWC-2011 best student paper award (2006), and of the NEWCOM++ distinguished achievement award 2008-2011.

Joakim Jaldén (S'03-M'08) received the M.Sc. and Ph.D. in electrical engineering from the Royal Institute of Technology (KTH), Stockholm, Sweden in 2002 and 2007 respectively. From July 2007 to June 2009 he held a post-doctoral research position at the Vienna University of Technology, Vienna, Austria. He also studied at Stanford University, CA, USA, from September 2000 to May 2002, and worked at ETH, Zurich, Switzerland, as a visiting researcher, from August to September, 2008. In July 2009 he joined the Signal Processing Lab within the School of Electrical Engineering at KTH, Stockholm, Sweden, as an Assistant Professor. He was an associate editor for the IEEE Communications Letters between 2009 and 2011, is as an associate editor for the IEEE Transactions in Signal Processing since 2012, and is a member of the IEEE Signal Processing for Communications and Networking Technical Committee (SPCOM-TC). For his work on MIMO communications, Joakim has been awarded the IEEE Signal Processing Society's 2006 Young Author Best Paper Award and the first price in the Student Paper Contest at the 2007 International Conference on Acoustics, Speech and Signal Processing (ICASSP). He is also a recipient of the Ingvar Carlsson Award issued in 2009 by the Swedish Foundation for Strategic Research, and the co-author of a best student paper at the 2012 International Symposium on Biomedical Imaging (ISBI).


T11 - An introduction to sparse stochastic processes
Subject Area: Fundamentals
Speaker: Micheal Unser (EPFL)

Summary
Sparse stochastic processes are continuous-domain processes that admit a parsimonious representation in some matched wavelet-like basis. Such models are relevant for image compression, compressed sensing, and, more generally, for the derivation of statistical algorithms for solving ill-posed inverse problems.

This tutorial provides an introduction to the extended family of sparse processes that are specified by a generic (non-Gaussian) innovation model or, equivalently, as solutions of linear stochastic differential equations driven by white Lévy noise. We provide a complete functional characterization of these processes and highlight some of their properties. The two leading threads that underlie the exposition are:
1) the statistical property of infinite divisibility, which induces two distinct types of behavior—Gaussian vs. sparse—at the exclusion of any other;
2) the structural link between linear stochastic processes and splines.
The formalism lends itself to the derivation of the transform-domain statistics of these processes and to the identification of “optimal” (ICA-like) representations. We also show that these models are applicable to the derivation of statistical algorithms for solving ill-posed inverse problems, including compressed sensing. The proposed formulation leads to a reinterpretation of popular sparsity-promoting processing schemes—such as total-variation denoising, LASSO, and wavelet shrinkage—as MAP estimators for specific types of sparse processes, but it also suggests alternative Bayesian recovery procedures that minimize the estimation error.

The lecture notes for the tutorial are available on the web at http://www.sparseprocesses.org

Biography

Michael Unser is Professor and Director of EPFL's Biomedical Imaging Group, Lausanne, Switzerland. His main research area is biomedical image processing. He has a strong interest in sampling theories, multiresolution algorithms, wavelets, the use of splines for image processing, and, more recently, stochastic processes. He has published about 250 journal papers on those topics. He is the leading author of “An introduction to sparse stochastic processes” to be published by Cambridge University Press. From 1985 to 1997, he was with the Biomedical Engineering and Instrumentation Program, National Institutes of Health, Bethesda USA, conducting research on bioimaging and heading the Image Processing Group. Dr. Unser is a fellow of the IEEE (1999), an EURASIP fellow (2009), and a member of the Swiss Academy of Engineering Sciences. He is the recipient of several international prizes including three IEEE-SPS Best Paper Awards and two Technical Achievement Awards from the IEEE (2008 SPS and EMBS 2010).


T12 - Factoring Tensors in the Cloud: A Tutorial on Big Tensor Data Analytics
Subject Area: Fundamentals
Speakers: Nicholas Sidiropoulos (University of Minnesota) and Evangelos Papalexakis (Carnegie Mellon University)

Summary
Tensors are data structures indexed by three or more indices – a generalization of matrices, which are datasets indexed by two indices (row, column). Tensor factorizations have already found many applications in signal processing (speech, audio, communications, radar, signal intelligence, machine learning) and well beyond, and they are becoming increasingly important – especially for analyzing big data, and tensors easily turn really big, e.g., 50 x 40 x 30 x 20 = 1.2 million. This tutorial has a two-fold purpose: Introduce the state-of-art in tensor decomposition research as it relates to big data analytics, and give participants a good idea of how the Hadoop/MapReduce paradigm of parallel and distributed computation works – and how it can be used to perform tensor factorization in the cloud. Besides Hadoop/MapReduce, we will also explore alternative means of parallelization. This tutorial will comprise four parts:
•    Basic tensor decomposition theory and methods, using easy-to-grasp examples and illustrations to reinforce familiar concepts that carry over from matrices to tensors, as well as what is special about tensors, with an emphasis on low-rank decomposition.
•    A tutorial look into recent developments in generalized sampling and multi-way compressed sensing strategies suitable for dealing with big tensors that cannot fit in main memory, or must reside in distributed storage.
•    A primer on the Hadoop/MapReduce model of parallel and distributed computation, first using a toy problem, then sketching how tensor decomposition can be mapped in a similar fashion.
•    Exciting applications in brain imaging and automatic language learning (in conjunction with the Read the Web project at CMU - http://rtw.ml.cmu.edu/rtw/), among others, will be interspersed throughout the tutorial, and also revisited in the end, to showcase the present day limits of parallel and distributed tensor computations. (This is joint work with Christos Faloutsos and the CMU team on Big Tensor Data)

Biographies

Nicholas Sidiropoulos (Fellow, IEEE) is currently a Professor in the Department of ECE at the University of Minnesota. He has over 15 years of experience in tensor decomposition and its applications, having been the first to introduce low-rank tensor decomposition to the signal processing society in the late 90’s. His research interests include topics in signal processing for communications, convex optimization, and cross-layer resource allocation for wireless networks His current research focuses primarily on signal and tensor analytics, with applications in cognitive radio, big data, and preference measurement. He received the NSF/CAREER award in 1998, and the IEEE Signal Processing Society (SPS) Best Paper Award in 2001, 2007, and 2011. He served as IEEE SPS Distinguished Lecturer (2008-2009), and as Chair of the IEEE Signal Processing for Communications and Networking Technical Committee (2007-2008). He served as Associate Editor for IEEE Transactions on Signal Processing (2000 - 2006), IEEE Signal Processing Letters (2000 - 2002), and on the editorial board of IEEE Signal Processing Magazine (2009-2011). He currently serves as Area Editor for IEEE Transactions on Signal Processing (2012 -), and as Associate Editor for Signal Processing. He received the 2010 IEEE Signal Processing Society Meritorious Service Award.

Evangelos Papalexakis is a Ph.D. student in the Computer Science Department at Carnegie Mellon University, working with Christos Faloutsos. Evangelos earned his Diploma and M.Sc. at TU Crete, Greece, where he worked with Nikos Sidiropoulos. Evangelos has considerable experience in both tensor decompositions and parallel and distributed computations in Hadoop. He has authored several papers in the area; he has published in the Transactions on Signal Processing and ICASSP, as well as ACM and top CS conferences in the area. At ease in both worlds, Evangelos understands what each community knows and how to bridge the gap between the two. Evangelos is also the liaison that connects the CMU and UMN groups in a joint NSF project on Big Tensor Data and its applications in automated web-based language learning and brain data mining.


T13 - Complex elliptically symmetric distributions and their applications in signal processing
Subject Area: Statistical Signal Processing
Speakers: Esa Ollila (Aalto University, Finland), David E. Tyler (Rutgers University) and Frederic Pascal (SUPELEC)

Summary
Complex elliptically symmetric (CES) distributions have been widely used in various engineering applications where non-Gaussian models are called for. In this tutorial, circularly symmetric CES distributions are surveyed, some new results are derived and their applications e.g., in radar and array signal processing are discussed and illustrated with theoretical and real-word examples, simulations and analysis of real radar data. A particular emphasis is put on maximum likelihood (ML) estimation of the scatter (covariance) matrix parameter.  General conditions for its existence and uniqueness, and for convergence of the iterative fixed point algorithm are discussed in detail. Specific ML-estimators for several CES distributions that are widely used in the signal processing literature are discussed in depth, including the complex t-distribution, K-distribution, the generalized Gaussian distribution and the recently proposed generalized inverse Gaussian distribution. Also the closely related angular central Gaussian distribution is discussed. A generalization of ML-estimators, the M-estimators of scatter matrix, are also discussed and asymptotic analysis is provided. The tutorial also contains new results on regularized (penalized) M-estimators of scatter, which are needed e.g., in low sample support scenarios.   Applications of CES distributions and the adaptive signal processors based on conventional and regularized ML- and M-estimators of the scatter matrix are illustrated in diverse applications such as radar detection problems and in array signal processing applications for Direction-of-Arrival (DOA) and beamforming in low sample support cases. Also graphical models and applications in image processing are discussed. Furthermore, experimental validation of the usefulness of CES distributions for modelling real radar data is given. The tutorial is partially based on our recent article (IEEE Trans. SP, vol. 60, no. 11, pp. 5597-5625) and to our other recent (also unpublished) works in this field.

Biographies

Esa Ollila (IEEE Member, 2003) received the M.Sc. degree in mathematics from the University of Oulu, in 1998, Ph.D. degree in statistics with honors from the University of Jyvaskyla, in 2002, and  the D.Sc. (Tech) degree with honors in signal processing from Aalto University, in 2010. From 2004 to 2007 he was a post-doctoral fellow of the Academy of Finland. He has also been a Senior Researcher and a Senior Lecturer at Aalto University and University of Oulu, respectively. Currently, from August 2010, he is appointed as an Academy Research Fellow of the Academy of Finland at the Department of Signal Processing and Acoustics, Aalto University, Finland. He is also adjunct Professor (statistics) of University of Oulu. During the Fall-term 2001 he was a Visiting Researcher with the Department of Statistics, Pennsylvania State University, State College, PA while the academic year 2010-2011 he spent as a Visiting Post-doctoral Research Associate with the Department of Electrical Engineering, Princeton University, Princeton, NJ. His research interests focus on theory and methods of statistical signal processing, blind source separation, complex-valued signal processing, array and radar signal processing and robust and non-parametric statistical methods.

David E. Tyler received the M.Sc. and Ph.D. degree  from the Department of Statistics, Princeton University,  in 1976 and 1979, respectively, and M.Sc. degree from the Department of  Mathematics and Statistics, University of Massachusetts. He is currently a Distinguished Professor in the Department of Statistics and Biostatistics at Rutgers University, where he has been on the faculty since 1983, serving as chair of the department from 1993-1996. From 1978-1979, he was an assistant professor in the Department of Statistics, University of Florida, and from 1979-1983, assistant professor in the Department of Applied Mathematics, Old Dominion University. He has held numerous visiting positions, including the University of Pittsburgh, Columbia University, the University of Toronto, the University of Leeds, the Swiss Institute of Technology, the University of Tampere and Princeton University. Dr. Tyler is a fellow of the Institute of Mathematical Statistics. He has served as an Associate Editor for various statistical journals, including the Annals of Statistics and the Journal of the Royal Statistical Society, as guest editor for special issues, and is a member of the steering committee of the International Conferences on Robust Statistics. His current research interests include robust statistics, multivariate analysis and spectral analysis of time series.

Frederic Pascal was born in Sallanches, France in 1979. He received the Master's degree ("Probabilities, Statistics and Applications: Signal, Image et Networks") with merit, in Applied Statistics from University Paris VII - Jussieu, Paris, France, in 2003. Then, he received the Ph.D. degree of Signal Processing, from University Paris X - Nanterre, advised by Pr. Philippe Forster entitled “Detection and Estimation in Compound Gaussian Noise" in 2006. This Ph.D. thesis was in collaboration with the French Aerospace Lab (ONERA), Palaiseau, France. From November 2006 to February 2008, he made a post-doctoral position in the Signal Processing and Information team of the laboratory SATIE (Systeme et Applications des Technologies de l'Information et de l'Energie), CNRS, Ecole Normale Superieure (ENS), Cachan, France. From march 2008 until December 2011, he was an Assistant Professor in the SONDRA laboratory, SUPELEC and from January 2012, he is an Associate Professor in SUPELEC. His research interests are estimation in statistical signal processing and radar detection.


T14 - Signal Processing for Finance, Economics and Marketing Modeling and Information Processing
Subject Area: Financial data analysis
Speakers: Xiao‐Ping (Steven) Zhang (Ryerson University), Fang Wang (Wilfrid Laurier University)

Summary
Financial time series is a major interest for signal processing researchers. Our signal processing (SP) researchers are often eager to apply signal processing techniques to stock price prediction or profitable trading strategy through analysis of price time series, i.e., conducting some forms of technical analysis, without knowing that technical analysis and most forms of stock price prediction are not well accepted in finance academia. Among the major finance theories, technical analysis contradicts the efficient-market hypothesis (EMH), a foundational theory in academic finance and economics research, but shares some common ground with behavioral economics theories. Lacking foundation knowledge of finance and economic theories constrains the potential of SP researchers in developing meaningful research in these areas. On the other hand, the demand and potential of applying SP techniques to finance, economics and marketing research are increasing. The tremendous increase of the amount of economic data in digital forms presents SP researchers huge opportunities. Many economics applications (including finance and marketing) beyond stock price prediction are available and research questions await answers. Indeed, many signal processing models and methods share common mathematical grounds with traditional econometric analysis but present different analytical aspects, and can therefore provide new tools for economic system modeling, analysis and information extraction for massive economic data. This tutorial intends to introduce the main stream foundational concepts and framework in finance, economics and marketing research, elaborate the relationships between traditional economic research paradigm and signal processing methodology, and help SP researchers identify relevant research directions. There are no special prerequisites for the audience. The tutorial is designed to present a refreshing signal processing perspective on finance, economicsand marketing research, and in-depth examples on SP applications in these fields. It will be valuable to SP researchers who are looking to broaden their knowledge beyond their current areas of expertise. The following major topics will be covered (i) Risk premium, the Capital Asset Pricing Model (CAPM) and its relationship to Wiener filter; (ii) Introduction on the EMH and behavioral economics theories; (iii) Basic econometrics models and relationship to signal processing models; (iv) Example applications to use signal processing models for marketing and financial economics modeling; (v) Potential signal processing research areas.

Biographies

Xiao-Ping (Steven) Zhang received B.S. and Ph.D. degrees from Tsinghua University, in 1992 and 1996, respectively, both in Electronic Engineering. He holds an MBA in Finance, Economics and Entrepreneurship with Honors from the University of Chicago Booth School of Business, Chicago, IL. Since Fall 2000, he has been with the Department of Electrical and Computer Engineering, Ryerson University, where he is now Professor, Director of Communication and Signal Processing Applications Laboratory (CASPAL). He is cross appointed to the Finance Department at the Ted Rogers School of Management at Ryerson University. Prior to joining Ryerson, he was a Senior DSP Engineer at SAM Technology, Inc., San Francisco. He held postdoctoral research and teaching positions at McMaster University, the Beckman Institute at UIUC, and the University of Texas, San Antonio. His research interests include signal processing and multimedia content analysis, sensor networks and electronic systems, pattern classification, and applications in finance, economics and marketing. He is a frequent consultant for hedge funds. He is cofounder and CEO for EidoSearch, an Ontario based company offering a search and analysis engine for financial data whose customers includes mainstream Wall Street banks and fund managers. Dr. Zhang is a registered Professional Engineer in Ontario, a Senior Member of IEEE and a member of Beta Gamma Sigma Honor Society. He served as guest editor for the Multimedia Tools and Applications, and the International Journal of Semantic Computing. He is currently an Associate Editor for IEEE Transactions on Signal Processing, IEEE Transactions on Multimedia, and IEEE Signal Processing Letters.

Fang Wang received Ph.D. in Management Information Systems from Michael G. DeGroote School of Business, McMaster University, and MBA in Finance from the University of Texas, San Antonio. Dr. Wang is an Associate Professor in Marketing at the School of Business & Economics, Wilfrid Laurier University. She held full-time management positions in the financial industry such as with HSBC. Her research interests include market response models, brand equity assessment, marketing data mining/analysis, information system management, long-term marketing productivity, big data and social media applications in marketing and finance.


T15 - Signal Processing in Power Line Communication Systems
Subject Area: Communications
Speaker: Andrea M. Tonello (University of Udine, Italy)

Summary
This tutorial focuses on signal processing techniques in the context of Power line communication (PLC) systems. It covers the application scenarios of PLC, channel characterization and modeling, signal processing at the physical layer, and the main aspects of the MAC layer with emphasis to scheduling and adaptation algorithms. The specific aspects of PLC that differentiate it from wireless will be emphasized showing that the channel poses significant challenges and opportunities to develop signal processing algorithms. As a starting point, an overview of the various application scenarios of PLC (such as in-home, in-vehicle, and smart grids) and a summary about the evolution of PLC technology will be provided. In reference to the Smart Grid, PLC is among the most interesting and important communication technology candidates since the grid is not only the information source but it also offers the infrastructure for the information delivery. The important topics of channel and noise modeling will be discussed. Up-to-date results about statistical channel modeling, MIMO channel modeling, and noise/disturbances modeling will be reported. Results of experimental channel characterization by different teams in different countries will be shown. Similarities and differences with the wireless channel behavior will be highlighted. The main challenges of physical layer design for both narrow-band (NB-PLC) and broad-band PLC (BB-PLC) to encompass the presence of channel attenuation and frequency selectivity, interference, and various noise sources will be addressed. In particular, a description will be offered about existing and emerging single carrier modulation approaches, filter bank modulation approaches (as OFDM, DWMT, FMT, cyclic block FMT) and ultra wide band techniques. We will show that advanced modulation techniques, combined with coding and smart resource allocation algorithms are capable to grant robust performance and coexistence with other technologies. We will also briefly discuss MAC protocols and signal processing for cooperative systems (as relaying) to enhance coverage and capacity in both home and smart grid PLC networks. Finally, an overview of the main standards will be offered covering both NB-PLC and broad-band BB-PLC. Their usage to deliver Smart Grid services (as automatic meter management, grid monitoring, vehicle-to-grid communication, demand side management, home networking for energy management) will be discussed.

Biography

Andrea M. Tonello
received the Doctor of Engineering degree in electrical engineering (1996) and the Doctor of Research degree in electronics and telecommunications (2003), both from the University of Padova, Italy. From 1997 to 2002, he was with Bell Labs – Lucent Technologies firstly as a Member of Technical Staff in the Advanced Wireless Technology Laboratory, Whippany, NJ. In 2001 he was promoted to Technical Manager and he was appointed Managing Director of the Bell Labs Italy division. In January 2003, he joined the University of Udine, Italy, where he is an Aggregate Professor and the founder of the Wireless and Power Line Communication Lab. He is also the founder and president of WiTiKee, a university spin-off company. Dr. Tonello received five best paper awards, the Lucent Bell Labs Recognition of Excellence Award (2003), the RAENG (UK) Distinguished Visiting Fellowship (2010) and the IEEE VTS Distinguished Lecturer Award (2012-15). He serves as an Associate Editor for the IEEE Transactions on Communications and IEEE Access. He is Vice-chair of the IEEE Technical Committee on PLC and the General Co-chair of IEEE SmartGridComm 2014, Venice, Italy.