Interference Mitigation and Collision Avoidance of Dynamic UAVBSs Network via Mobility Control: A Game Theoretic Approach

Mobility control of Unmanned Aerial Vehicle Base Stations (UAVBSs) can avoid collision and improve the power efficiency and coverage of the wireless network. In this work, UAVBS mobility control is formulated as an exact potential game. Three algorithms are proposed to solve this problem under different connectivity and complexity scenarios. In the first scenario on board computation and power may be limited due to other functions. Under this scenario, the UAVBSs-Better Direction Control (UAVBSs-BDC) algorithm works iteratively based only on the UAV utility function with linear time to directly optimize the action selection based on the UAVBS’s utility. The Utility-Driven Partial Synchronous Learning (UDPSL) algorithm speeds up convergence by using a learning algorithm. This algorithm is seen to increase the incidence of collision when UAVBSs are located close together and require an additional collision avoidance mechanism. The Neighbor Responsive Adaptive-Partial Synchronous Learning (NRA-PSL) algorithm controls the UAVBS’s trajectory via conditioned response to its neighbor UAVBSs to select the action that guides the UAVBS towards a better direction. This algorithm requires additional information about the interference posed by neighbor UAVBS and their location in the cell, which allows it to design a better trajectory which converges faster to the optimal placement of UAVBSs in the cell.