Abstract: Wireless communication systems have traditionally been designed to operate under the channel conditions provided by nature, with the Shannon capacity being the ultimate limit. The advent of reconfigurable intelligent surfaces (RIS) changes the status quo by adding the ability to control wave propagation, thereby modifying the capacity. A RIS is an array of reflecting elements with properties that can be tuned to synthesize the reflection behavior of a differently shaped object. The larger the surface is, the greater performance improvements are possible. The gains are achieved by adapting the RIS configuration to the channel coefficients from the transmitter to the receiver via the RIS, which must be estimated in practice. The estimation challenge grows with the number of reflecting elements, in terms of the signal resources required for pilot signaling. The problem differs from the classical estimation problems in multi-antenna communications because the RIS is blind and substantially larger. Hence, the estimation dimensionality might be the showstopper for practical RIS deployments. In this talk, we will look at how array signal processing theory provides suitable tools to describe and analyze RIS-aided communication systems. We will explore how these tools enable us to model the physical channels and discover how to simplify the channel estimation problem by exploiting fundamental properties, such as the array geometry, array responses, and channel geometry. It turns out that, under the right circumstances, array processing methodology can solve the estimation challenge. The talk will end with an outlook on unsolved showstoppers for RIS technology.

Bio: Emil Björnson is a Professor of Wireless Communication at the KTH Royal Institute of Technology, Stockholm, Sweden. He is an IEEE Fellow, Digital Futures Fellow, and Wallenberg Academy Fellow. He has a podcast and YouTube channel called Wireless Future. His research focuses on multi-antenna communications and radio resource management, using methods from communication theory, signal processing, and machine learning. He has authored four textbooks and published a large amount of simulation code. 

He has received the 2018 and 2022 IEEE Marconi Prize Paper Awards in Wireless Communications, the 2019 EURASIP Early Career Award, the 2019 IEEE Communications Society Fred W. Ellersick Prize, the 2019 IEEE Signal Processing Magazine Best Column Award, the 2020 Pierre-Simon Laplace Early Career Technical Achievement Award, the 2020 CTTC Early Achievement Award, the 2021 IEEE ComSoc RCC Early Achievement Award, and the 2023 IEEE Communications Society Outstanding Paper Award. His work has also received six Best Paper Awards at conferences.

Abstract:  The first part of the talk will serve as a mini tutorial, highlighting the key ideas behind tensor-based data analysis. We will explain that the Canonical Polyadic Decomposition (CPD) and the Block Term Decomposition (BTD) of higher-order tensors are fundamental tools for blind signal separation and latent variable analysis, “beyond matrix techniques”. 

In the second part of the talk, we will take the step from tensor-based data analysis to tensor-based pattern recognition, and from the decomposition of a single tensor to the assessment of the similarity between components of different tensors.

Assessing similarity is a key task in pattern recognition and machine learning. We will show that tensors also provide fundamentally new possibilities for blind similarity assessment and latent variable matching, “beyond matrix pair techniques”. Moreover, under mild conditions, the assessment of similarity can be done by conventional linear algebra.

The results will be illustrated with applications.

Bio:  Lieven De Lathauwer is Full Professor at KU Leuven, Belgium, affiliated with both the Group Science, Engineering and Technology of Kulak (where he is academic coordinator Data Science and AI), and with the STADIUS division of the Electrical Engineering Department (ESAT). His research concerns the development of tensor tools for mathematical and electrical engineering. This interdisciplinary work centers on the following axes: (i) algebraic foundations, (ii) numerical algorithms, (iii) generic methods for signal processing, data analysis, system modelling and machine learning, and (iv)concrete applications in (biomedical) signal processing, material sciences, telecommunication, and other fields. Algorithms have been made available as Tensorlab (www.tensorlab.net). Further, Tensorlab+ is a reproducible research repository for tensor computations (www.tensorlabplus.net). He (co-)chaired Workshops on Tensor Decompositions and Applications (TDA 2005, 2010, 2016). He was a co-recipient of the IEEE SPS Signal Processing Magazine Best Paper Award in 2018 and a co-recipient of the IEEE SPS Donald G. Fink overview paper award in 2023. He is Fellow of IEEE (2015), Fellow of SIAM (2017) and Fellow of EURASIP (2019).