Reinforcement Learning Placement Algorithm for Optimization of UAV Network in Wireless Communication

Unmanned aerial vehicles (UAVs) or drones have attracted much attention in wireless communication networks because of their agility, unique flexibility, low cost of implementation, and the high strength of the line-of-sight (LoS) channel. They are widely used in different scenarios. In many environments with complex geographical conditions or in situations where areas are affected by natural disasters, UAVs can be used as base stations (BSs) for downlink ground users. The article proposes a communication system using multiple UAV-mounted BSs to improve coverage rate and minimize the number of required UAVs. The problem is formulated as a mixed-integer programming problem with constraints on the quality of service (QoS) and serviceability of each UAV. A three-step method is developed to solve the problem, which includes deriving the maximum service radius of UAVs using the Karush-Kuhn-Tucker (KKT) method, minimizing the number of required UAVs using reinforcement learning (RL) algorithm, and designing the three-dimensional (3D) position and frequency band of each UAV to increase signal power and reduce interference. The simulation results show that the RLP algorithm outperforms other algorithms in terms of coverage rate, user clustering, increased signal, reduced interference, and processing time required to find the optimal solution.