Channel-Adaptive Robust Resource Allocation for Highly Reliable IRS-Assisted V2X Communications

This paper investigates robust resource allocation in intelligent reflecting surface (IRS)-assisted vehicle-to-everything (V2X) networks under unknown channel state information (CSI) uncertainties induced by Doppler shifts and high vehicle mobility. Unlike most existing IRS-assisted V2X studies that assume perfect or statistically known CSI, we address the more realistic yet under-explored case of distribution-free imperfect CSI. We formulate a joint optimization problem to maximize the sum capacity of vehicle-to-infrastructure (V2I) links while satisfying probabilistic quality-of-service (QoS) constraints for safety critical vehicle-to-vehicle (V2V) communications. The problem jointly optimizes transmit power, multi-user detection (MUD) matrices, spectrum reuse patterns, and IRS phase shifts. To tackle this, a novel adaptive channel-based alternating optimization (ACAO) method is developed by first decomposing the problem into smaller sub-problems such as power optimization, MUD matrices optimization and IRS phase shifts, and then iteratively optimizing these sub-problems using Block Coordinate Descent (BCD) approach to achieve near-optimal resource allocation. Key contributions include an efficient self-learning approach for power allocation and a tailored phase shift optimization technique that effectively adapt to unknown CSI uncertainties, significantly enhancing the performance of V2X communications and reducing the computation complexity. Simulation experiments confirm the efficacy of the presented solutions in improving the overall capacity compared to benchmarks.