PhD Seminar Notice: Optimizing RIS-assisted Wireless Communications

Candidate: Mahmoud Abouamer
Date: September 20, 2024
Time: 10:00 AM
Place: EIT 3145
Supervisor(s): Mitran, Patrick

Abstract:
Future wireless technologies are expected to cater to the growing demand for connectivity while adhering to strict quality-of-service (QoS) requirements such as high spectrum and power efficiency, low latency, and extended coverage. Reconfigurable intelligent surfaces (RIS), also known as intelligent reflecting surfaces (IRS), offer a promising solution. An RIS consists of a 2D array of a large number of nearly-passive reflecting elements that can dynamically shape incident electromagnetic waves. This nearly-passive design allows RIS to operate in full-duplex with no self-interference and with low hardware complexity. By integrating an RIS into a communication system, new degrees of freedom can be leveraged to collectively modify the incident signal paths and engineer the wireless propagation environment to assist signal transmission. This research addresses these challenges by developing RIS configuration schemes while considering configuration complexity and overhead.

First, we focus on RIS design in multi-user frequency-division-duplexing (FDD) and time-division-duplexing (TDD) systems. A joint uplink-downlink RIS design is proposed, where the same IRS configuration assists both uplink and downlink transmissions. A weighted-sum problem is formulated and optimized using a block-coordinate descent algorithm to compute the trade-off between uplink and downlink rates. Numerical simulations explore the resulting uplink-downlink trade-off regions.

In the second phase, we develop a learning-based framework that configures RIS with limited channel estimation overhead while addressing varying service priorities and fairness. An adaptive learning problem is formulated to learn beamforming configuration schemes that optimize the weighted sum-rate (WSR) in multi-user RIS systems using pilot information. This problem is analyzed, and  a novel hypernetwork-based beamforming (HNB) framework is proposed to adaptively optimize RIS and beamforming configurations without retraining as user weights change.

The third phase aims to relax hardware constraints and enable RIS implementation with diverse technologies while accurately modeling its reflection behavior. Specifically, we consider configuring an RIS using discrete phase shifts that exhibit practical non-linear  amplitude responses. Subsequently, a low-complexity configuration framework is proposed. Future work aims to gain insights into the performance of the proposed low-complexity algorithm in configuring a discrete RIS by investigating the structure of the configuration problem and obtaining performance guarantees under proposed configuration framework.