Genetic Algorithm Combined with Ray Tracer for Optimizing Cell-Free mMIMO Topology in a Confined Environment

Abstract

This paper proposes a customized genetic algorithm (GA) to generate the optimal cell-free topology for multi-user massive MIMO (mMIMO) in a confined environment. As far as we know, it is beyond the literature and is the first attempt to apply GA in optimizing the base station (BS) antenna placement for cell-free mMIMO. The BS antennas' placement is encoded with an adjusted binary matrix representation, which is straightforward for the subsequent genetic operations. The explored candidates by GA can evolve beyond the parents, where the fitness of individuals is evaluated dynamically via a ray tracer channel simulator. Accelerated by a warm start strategy and elitist replacement, the proposed customized GA provides near-optimal results in experiments, applicable to generic environment with multiple mobile users and different signal-to-noise ratios.