A Resource Allocation Policy for Downlink Communication in Distributed IRS Aided Multiple-Input Single-Output Systems

As a technology for 6G wireless communications, Intelligent Reflecting Surfaces (IRSs) are considered as a promising solution to boost the network capacity, spectrum and coverage in multiusers' downlink communication systems. The users in blockage and cell edge areas can utilize this technology for data transfer purpose. In this paper, a machine learning-based policy optimization for downlink communication in distributed IRS aided multiple-input single-output (MISO) systems is proposed. Three categories of users are considered, namely, users who can utilize only the direct links, blockage area users who can utilize only the IRS links, and cell edge or poor link quality of users who can utilize both the direct and IRS links. The sum rate maximization problem is formulated to derive the optimal policy (i.e. communication link, IRS selection, power allocation and reflection coefficients) for those users, considering the IRS selection, link quality, power allocation and IRS reflection constraints. The proposed methods to achieve the optimal policy include reinforcement learning-based model with binary decision tree-based user categories, maximum posterior probability-based IRS selection, fractional programming method-based power and IRS coefficient allocation, and value function-based policy optimization. Through simulations, the sum data rate and energy efficiency performances of different categories of users are obtained and discussed.