Forum Numerica - A Tensor Perspective to Large-Scale MIMO Array Processing


Federal University of Ceará, Brasil



Video / Presentation

To watch in full screen mode, start the video and click on the "UNS POD" logo


Tensors are multidimensional arrays that can be regarded as generalizations of matrices to higher orders (>2). In particular, tensor decompositions are powerful tools to solve a wide variety of applications, including wireless communications, blind source separation, biomedical signal processing, and data analysis. In this talk, we discuss the intertwining between tensor decompositions and large-scale MIMO array processing. The primary motivation lies in the multidimensional nature of the wireless channel that manifests itself in different physical domains (space, time, frequency, polarization, etc.), which translates into tensor structures for the parametric representation of the channel. In addition to its multidimensional structure, the sparse nature of the MIMO channel becomes more relevant in communication systems operating in the millimeter wave band. By jointly exploiting the sparse and multidimensional structures of the millimeter wave (mmWave) MIMO wireless channel, and in some cases, its low-rank property, computationally efficient tensor-based algorithms for channel estimation, hybrid precoding and spatial filtering can be derived. Recent progress in this field will be discussed from a tensor perspective and some challenges and research directions will be raised.

About the speaker

André L. F. de Almeida is currently an Associate Professor with the Department of Teleinformatics Engineering of the Federal University of Ceará. He received the double Ph.D. degree in Sciences and Teleinformatics Engineering from the University of Nice, Sophia Antipolis, France, and the Federal University of Ceará, Fortaleza, Brazil, in 2007. During fall 2002, he was a visiting researcher at Ericsson Research Labs, Stockholm, Sweden. From 2007 to 2008, he held a one-year teaching position at University of Nice Sophia Antipolis, France. In 2008, he was awarded a CAPES/COFECUB research fellowship with the I3S Laboratory, CNRS, France. He was awarded multiples times visiting professor positions at the University of Nice Sophia-Antipolis, France. He has over 60 journal articles published and accepted, 100 conference papers and 5 book chapters. He served as an Associate Editor for the IEEE Transactions on Signal Processing (2012-2016). He currently serves as an Associate Editor for the IEEE Signal Processing Letters. Dr. Almeida is a member of the Sensor Array and Multichannel (SAM) Technical Committee of the IEEE Signal Processing Society (SPS) and a member of the EURASIP Signal Processing for Multi-Sensor Systems Special Area Team (SAT-SPMuS). He was the general co-chair of the IEEE CAMSAP'2017 workshop, and served as the Technical Co-Chair of the Symposium on "Tensor Methods for Signal Processing and Machine Learning" at GlobalSIP'2018. He also serves as the technical co-chair of IEEE SAM 2020 workshop, Hangzhou, China. He is a research fellow of the CNPq (the Brazilian National Council for Scientific and Technological Development) and an elected affiliate member of the Brazilian Academy of Sciences (2018-2022). He is a Senior Member of the IEEE. His current research has a focus on tensor methods and multilinear algebra with applications to communications and signal processing.