2020/10/09 by George C. Alexandropoulos, Sumudu Samarakoon, Alexandropoulos, George C. +5 · 3 citations
Engineering · #Advanced Wireless Communication Technologies #Emerging Technologies (cs.ET) #FOS: Computer and information sciences #Indoor and Outdoor Localization Technologies #Information Theory (cs.IT) #Machine Learning (cs.LG) #Underwater Vehicles and Communication Systems
paper · pdf · doi:10.48550/arxiv.2010.04376
openalex publication_date 2020/10/09 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Reconfigurable Intelligent Surfaces (RISs) are recently gaining remarkable\nattention as a low-cost, hardware-efficient, and highly scalable technology\ncapable of offering dynamic control of electro-magnetic wave propagation. Their\nenvisioned dense deployment over various obstacles of the, otherwise passive,\nwireless communication environment has been considered as a revolutionary means\nto transform them into network entities with reconfigurable properties,\nproviding increased environmental intelligence for diverse communication\nobjectives. One of the major challenges with RIS-empowered wireless\ncommunications is the low-overhead dynamic configuration of multiple RISs,\nwhich according to the current hardware designs have very limited computing and\nstorage capabilities. In this paper, we consider a typical communication pair\nbetween two nodes that is assisted by a plurality of RISs, and devise\nlow-complexity supervised learning approaches for the RISs' phase\nconfigurations. By assuming common tunable phases in groups of each RIS's unit\nelements, we present multi-layer perceptron Neural Network (NN) architectures\nthat can be trained either with positioning values or the instantaneous channel\ncoefficients. We investigate centralized and individual training of the RISs,\nas well as their federation, and assess their computational requirements. Our\nsimulation results, including comparisons with the optimal phase configuration\nscheme, showcase the benefits of adopting individual NNs at RISs for the link\nbudget performance boosting.\n