IRS-Based DOA Estimation in C-RAN ISAC System

This paper investigates an active intelligent reflecting surface (IRS) assisted integrated sensing and communications (ISAC) system in a cloud radio access network (C-RAN). In particular, we focus on the IRS assisted target sensing for the blind area, wherein multiple IRSs are deployed to establish controllable reflective links between the targets and the sensing remote radio heads (RRHs). A novel IRS-assisted location-aware direction-of-arrival (DOA) estimation scheme is proposed, where the BBU (baseband unit) pool simultaneously recovers the DOA from each target to each IRS using the RRH received signals. In general, the channel knowledge between the IRSs and RRHs is hard to acquire, hence we utilize the location information of IRSs and RRHs instead. Specifically, we transform the DOA estimation problem into a mixed one-dimensional and two-dimensional atomic norm minimization (ANM) problem, by which the DOAs can be efficiently extracted. Moreover, a theoretical Cramér-Rao lower bound (CRLB) on the DOA estimation error is also derived to evaluate the performance of the proposed scheme. Finally, simulation results illustrate the effectiveness for DOA estimation by the proposed scheme.