Low Dynamic Range for RIS-aided Bistatic Integrated Sensing and Communication

TL;DR

This paper proposes an optimization framework for RIS-aided integrated sensing and communication systems to minimize path interference, enhancing system performance through advanced beamforming and phase shift design.

Contribution

It introduces a novel non-convex optimization approach using BCCD and Riemannian conjugate gradient methods for interference minimization in RIS-assisted ISAC systems.

Findings

Effective interference reduction demonstrated in simulations

Proposed algorithms outperform benchmark methods

Enhanced system performance with optimized beamforming and phase shifts

Abstract

The following paper presents a reconfigurable intelligent surface (RIS)-aided integrated sensing and communication (ISAC) system model scenario, where a base station communicates with a user, and a bi-static sensing unit, i.e. the passive radar (PR), senses targets using downlink signals. Given that the RIS aids with communication and sensing tasks, this paper introduces new interfering paths that can overwhelm the PR with unnecessarily high power, namely the path interference (PI), \textit{which is itself a combination of two interfering paths, the direct path interference (DPI) and the reflected path interference (RPI)}. For this, we formulate an optimization framework that allows the system to carry on with its ISAC tasks, through analog space-time beamforming at the sensing unit, in collaboration with RIS phase shift and statistical transmit covariance matrix optimization, while minimizing the PI power. As the proposed optimization problem is non-convex, we tailor a block-cyclic coordinate descent (BCCD) method to decouple the non-convex sub-problem from the convex one. A Riemannian conjugate gradient method is devised to generate the RIS and PR space-time beamforming phase shifts per BCCD iteration, while the convex sub-problem is solved via off-the-shelf solvers. Simulation results demonstrate the effectiveness of the proposed solver when compared with benchmarking ones.