Costas Sarris of University of Toronto

Reconfigurable intelligent surfaces (RISs) can redirect incident waves to selected directions in a wireless communication channel, to improve coverage bypassing obstructing objects. This is a promising technology for future wireless networks, particularly those operating at millimeter wave bands. In practice, several factors affect the performance of an RIS-enabled communication channel. For example, both the transmitter and the RIS radiate waves according to a pattern that includes a main lobe along with sidelobes, leading to enhanced multipath propagation. Moreover, most existing propagation models represent RISs as a collection of independently scattering unit cells, disregarding their mutual coupling that is known to be an important factor in the collective scattering behavior of these surfaces. On the other hand, full-wave methods offer high accuracy, but their computational cost is prohibitive for realistic radio environments.

This presentation introduces a hybrid approach that enables the accurate analysis of RIS-enabled communication channels in rich scattering, realistic environments. Our hybrid method combines the full-wave analysis of the RIS as a diffuse scatterer (including the mutual coupling of unit cells and edge effects), with the computational efficiency of ray-tracing to obtain accurate field predictions in the near and far field region of the RIS at multiple states, accounting for multipath propagation paths towards and from the RIS. A convolutional neural network is presented to efficiently compute scattered fields from a digital RIS, whose cells can be set to discrete states. Prof. Costas Sarris presents validation studies based on reference full-wave analysis of relatively small problems and on an extensive indoor measurement campaign. Finally, they discuss the practical use of their comprehensive propagation model to the site-specific analysis, design, and optimization of RIS structures and present a reinforcement-learning strategy for the fast optimization of RISs to meet specific pattern objectives.

Biography:

Costas Sarris received a Ph.D.  in Electrical Engineering and a M.Sc. in Applied Mathematics from the University of Michigan-Ann Arbor. He is a Professor with the Department of Electrical and Computer Engineering, University of Toronto. His research area is computational electromagnetics, with an emphasis on time-domain modeling. He also works on physics-based wireless propagation models (with full-wave, asymptotic, and hybrid techniques), uncertainty quantification, and scientific machine learning.

Dr. Sarris is an IEEE Fellow and a Distinguished Lecturer of the IEEE Antennas and Propagation Society for 2024-2026. He was a recipient of the 2021 Premium Award for Best Paper in IET Microwaves, Antennas & Propagation, and the IEEE MTT-S Outstanding Young Engineer Award in 2013. He was the TPC Chair of the 2015 IEEE AP-S International Symposium on Antennas and Propagation and the CNC/USNC Joint Meeting, the 2019 and 2023 MTT-S Numerical Electromagnetics, Multiphysics and Optimization (NEMO) Conference, the TPC Vice-Chair of the 2012 IEEE MTT-S International Microwave Symposium, and the Chair of the MTT-S Technical Committee on Field Theory and Numerical Electromagnetics (2018–2020). In 2019-2024, he was the Editor-in-Chief of the IEEE JOURNAL ON MULTISCALE AND MULTIPHYSICS COMPUTATIONAL TECHNIQUES.