Trajectory optimization of UAV-IRS assisted 6G THz network using deep reinforcement learning approach

Area of Science:

• Wireless Communication Engineering
• Optimization Algorithms
• Artificial Intelligence in Networks

Background:

• Terahertz (THz) communication promises ultra-high data rates and low latency for future 6G networks.
• Unmanned Aerial Vehicles (UAVs) integrated with Intelligent Reconfigurable Surfaces (IRS) are key enablers for 6G.
• Balancing mission completion time and energy efficiency is critical for UAV-IRS systems in THz networks.

Purpose of the Study:

• To address the challenge of minimizing mission completion time for UAV-IRS systems in THz networks.
• To optimize UAV-IRS trajectories considering energy consumption constraints.
• To develop an efficient algorithm for solving the NP-hard trajectory planning problem.

Main Methods:

• Formulation of the UAV-IRS trajectory planning as a non-convex optimization problem.
• Application of a Deep Q-Network (DQN) reinforcement learning algorithm to overcome conventional technique limitations.
• Low-complexity optimization algorithm design suitable for THz communication networks.

Main Results:

• The proposed DQN-based algorithm effectively optimizes UAV-IRS trajectories.
• The algorithm successfully minimizes mission completion time under energy constraints.
• Simulation results demonstrate superior performance compared to existing benchmark schemes.

Conclusions:

• The DQN reinforcement learning approach provides an efficient solution for UAV-IRS trajectory planning in THz networks.
• This method balances mission efficiency and energy consumption, crucial for 6G applications.
• The proposed algorithm offers a practical and high-performing solution for complex wireless communication scenarios.