Energy-Efficient UAV Movement Control for Fair Communication Coverage: A Deep Reinforcement Learning Approach.

Area of Science:

• Wireless Communication Networks
• Robotics and Control Systems
• Distributed Artificial Intelligence

Background:

• Unmanned Aerial Vehicles (UAVs) offer mobility and flexibility as base stations to enhance wireless communication quality and coverage.
• Challenges in UAV deployment include limited energy, short communication range, and regulatory constraints, necessitating distributed solutions for dynamic environments.
• Existing methods struggle with large state spaces and complex interactions among multiple UAVs.

Purpose of the Study:

• To develop a novel distributed control solution for optimizing UAV placement in wireless networks.
• To improve communication coverage score, minimize energy consumption, and ensure high fairness among ground users.
• To address the complexities of multi-UAV interactions and dynamic environmental conditions.

Main Methods:

• Introduced a state-based game with actor-critic (SBG-AC) algorithm for distributed UAV control.
• Modeled the SBG-AC algorithm using a state-based potential game to simplify complex interactions.
• Integrated SBG-AC with an actor-critic algorithm to ensure convergence and enable learning in dynamic environments.

Main Results:

• The SBG-AC algorithm demonstrated superior performance compared to distributed Deep Reinforcement Learning (DRL) and DRL-EC3.
• Achieved significant improvements in fairness, coverage score, and energy efficiency.
• Effectively managed distributed control and learning capabilities for UAVs in dynamic scenarios.

Conclusions:

• The proposed SBG-AC algorithm provides an effective distributed solution for UAV-assisted wireless communication.
• SBG-AC enhances network performance metrics including coverage, fairness, and energy efficiency.
• This approach offers a promising direction for intelligent UAV deployment in future communication systems.