Abstract
Reconfigurable intelligent surface (RIS) and unmanned aerial vehicle (UAV) are anticipated as talented technologies to extend the range of millimeter wave (mmWave) communications. In this letter, a UAV equipped with RIS (UAV-RIS) is used to assist mmWave base station (BS) in covering users in hotspot areas. In this context, UAV should cover several high-capacity hotspots while minimizing its flying/hovering energy consumptions. Energy-aware multi-armed bandit (EA-MAB) algorithm is proposed as an effective online learning tool to handle this problem efficiently. By which, the UAV acts as the player trying to maximize its achievable rate, i.e., the reward, over selecting different hotspots in its trajectory, i.e., the arms of the bandit game. This is done while minimizing the energy/cost of the UAV flight from one hotspot to another over the time span of its battery life. Numerical analysis confirms the superior performance of the proposed EA-MAB algorithm over benchmarks.