Aerial-IRS-Assisted Load Balancing In Downlink Networks

This work suggests a joint optimization of the aerial intelligent reflecting surface (AIRS) placement, passive beamforming, and base station (BS) association to improve the overall data throughput and fairness across downlink heterogeneous cellular networks. Differing from the related works in the literature that just seek to maximize signal-to-interference-plus-noise ratio (SINR), the paper takes into account the load balancing between macrocells and small cells. The resulting joint optimization problem is mixed continuous-discrete and has a highly bumpy landscape, so the traditional (sub)gradient-based tools are not suited. We propose a model-free approach based on adaptive particle swarm optimization (APSO) and blind beamforming, which recovers the solution from random explorations of the solution space. Simulations show that the proposed algorithm enables balanced traffic for the coexisting macro and small cells, and thereby achieves a higher network utility than the benchmark methods.