Deep reinforcement learning based rate enhancement scheme for RIS assisted mobile users underlaying UAV

Abstract

The fifth generation (5G) network enabled communication between devices has emerged as a state-of-the-art technology. In the era of proliferating smart devices and intelligent wireless communication networks, Reflecting Intelligent Surfaces (RIS) and Unpiloted Air Vehicles (UAV) duplet has turn out to be a trustworthy, lucrative and handy solution for various appearing real world communication issues. This article pitches into the downlink UAV communication empowered by RIS, where UAV communicates with Mobile Instruments (MI) via RIS patches installed at a tall tower. Considering the attributes like transmitted power and UAV trajectory, Deep Reinforcement Learning (DRL) based approach is recommended to maximize the overall Sum-rate. In present scenario, DRL technology has popped up as a commanding tool that allows a network to regulate itself in order to deliver optimum solution. In this article, we have proposed a novel viewpoint evolved from Deep Deterministic Policy Gradient (D-DPG) Algorithm specifically Shared Deep Deterministic Policy Gradient (SD-DPG) algorithm for downlink UAV-MI power allocation and trajectory optimization problem. Numerical outcomes manifest that our model, concerned to maximizing sum-rate, outperformed other DRL based method DD-DPG by at least 30% and D-DPG by approximately 3 folds together with optimizing power, phase-shift and UAV trajectory.

Keywords

• 5G network
• RIS
• UAV
• DRL
• Power allocation
• Sum-rate