Deep Reinforcement Learning Explores EH-RIS for Spectrum-Efficient Drone Communication in 6G

Abstract

Reconfigurable intelligent surfaces (RISs) have emerged as a groundbreaking technology, revolutionizing wireless networks with enhanced spectrum and energy efficiency (EE). When integrated with drones, the combination offers ubiquitous deployment services in communication-constrained areas. However, the limited battery life of drones hampers their performance. To address this, we introduce an innovative energy harvesting (EH), that is, EH-RIS. EH-RIS strategically divides passive reflection arrays across geometric space, improving EH and information transformation (IT). Employing a meticulous, exhaustive search algorithm, the resources of the drone-RIS system are dynamically allocated across time and space to maximize harvested energy while ensuring optimal communication quality. Deep reinforcement learning (DRL) is employed to investigate drone-RIS performance by intelligently allocating resources for EH and signal reflection. The results demonstrate the effectiveness of the DRL-based EH-RIS simultaneous wireless information and power transfer (SWIPT) system, demonstrating enhanced drone-RIS spectrum-efficient communication capabilities. Our investigation is summarized in unleashing potential, which shows how DRL and EH-RIS work together to optimize drone-RIS for next-generation wireless networks.