Trajectory Optimization of UAV-IRS Assisted 6G THz Network Using Deep Reinforcement Learning Approach

Abstract Terahertz (THz) wireless communication is a promising technology that will enable ultra-high data rate, and very low latency for future wireless communications. Intelligent Reconfigurable Surfaces (IRS) aiding Unmanned Aerial Vehicle (UAV) are two essential technologies play a pivotal role in balancing the demands of Sixth-Generation (6G) wireless networks. In practical scenarios, mission completion time and energy consumption serve as crucial benchmarks for assessing the efficiency of UAV-IRS enabled THz communication. Achieving a swift mission completion requires UAV-IRS to fly at maximum speed above the ground users it serves. However, this results in higher energy consumption. To address the challenge, this paper studies UAV-IRS trajectory planning problems in THz networks. The problem is formulated as an optimization problem aiming to minimize UAVs-IRS mission completion time by optimizing the UAV-IRS trajectory, considering the energy consumption constraint for UAVs-IRS. Proposed optimization algorithm, with low complexity, is well-suited for applications in THz communication networks. This problem is non-convex, optimization problem that is NP-hard and presents challenges for conventional optimization techniques. To overcome this, we proposed Deep Q-Network (DQN) reinforcement learning algorithm to enhance performance. Simulation results show that our proposed algorithm achieves performance compared to benchmark schemes.