Use of unmanned aerial vehicles (UAVs) to provide on-demand communications has been receiving growing interest, especially for use in remote and hard-to-reach areas. Also, the use of light-emitting diode -based lighting in UAVs has opened opportunities for data transmission through visible-light communications. To manage multi-user interference while avoiding complex handover procedures, we consider the use of zero forcing (ZF) pre-coding. Since the performance of ZF pre-coding depends on the correlation between channel gains of users, we propose in this paper to reduce it by means of location optimization of UAVs. More specifically, we use particle swarm optimization with the objective of maximizing the overall achievable network throughput. Furthermore, to relax the optimization requirements at UAVs, we investigate the case when the optimization is performed at a specific rate under different mobility conditions.