Long-Term Energy Consumption Minimization Based on UAV Joint Content Fetching and Trajectory Design

Area of Science:

• Wireless Communication Networks
• Optimization Theory
• Artificial Intelligence

Background:

• Unmanned Aerial Vehicles (UAVs) offer potential for improved content delivery in wireless networks.
• Efficient content fetching and energy management are critical challenges in multi-UAV systems.
• Existing methods struggle with the complexity of joint optimization problems.

Purpose of the Study:

• To minimize energy consumption in multi-UAV networks by jointly optimizing UAV trajectory, power allocation, content fetching, and content placement.
• To address the challenges of a complex mixed-integer nonlinear programming (MINLP) problem.
• To develop a novel, efficient solution for stochastic optimization in UAV-aided content delivery.

Main Methods:

• Formulated a constrained optimization problem for joint design parameters.
• Transformed the problem into a semi-Markov decision process (SMDP).
• Developed an option-based hierarchical deep reinforcement learning (OHDRL) technique, defining low-level actions (trajectory, power) and high-level options (content placement, fetching).

Main Results:

• The proposed OHDRL approach effectively solves the joint optimization problem.
• Numerical results demonstrate more consistent learning performance compared to existing techniques.
• The method achieves significant reductions in system energy consumption.

Conclusions:

• OHDRL provides an effective framework for optimizing complex UAV-aided content delivery systems.
• The developed approach balances performance gains with energy efficiency.
• This research offers a promising direction for future UAV network design and resource management.