Joint Resource Allocation and UAV Trajectory Design for D2D-Assisted Energy-Efficient Air-Ground Integrated Caching Network

Leveraging unmanned aerial vehicle (UAV) and high altitude platform station (HAPS) to cache popular content, and combining with D2D communication to build the D2D-assisted air-ground integrated caching network (AGICN-D2D), is an effective way to meet the increasing data demand and relieve the peak pressure of future network. However, due to the limited on-board energy storage of UAVs and terminals, optimizing resource allocation and UAV trajectory to improve system energy efficiency (EE) becomes a major issue. Moreover, due to the high dynamic nature of the AGICN-D2D, it often falls into the dilemma of information uncertainty and curse of dimensionality. To overcome the above challenges, a downlink AGICN-D2D model consisting of HAPS, UAVs, and users is intended, where user devices can cache popular content and assist data transmission through D2D links. Then, the joint content cache probability, resource allocation and UAV trajectory plan issue is investigated for maximizing system EE. As it is an NP-hard issue coupled with content caching probability, power control, bandwidth assignment and trajectory plan, we decompose it into two subproblems. Firstly, for the content caching probability sub-problem, AVOA-based Cache Probability Optimization (ACPO) algorithm is developed to realize the optimization of content caching probability. Secondly, for the power control, bandwidth allocation and trajectory design sub-problem, a multi-agent deep deterministic policy gradient (MADDPG)-based Resource allocation and Flight trajectory Optimization (MRFO) algorithm is developed to maximize the system EE. Finally, Joint Content caching probability, Resource allocation and Flight trajectory Optimization (JCRFO) algorithm is introduced. We conduct extensive experiments to compare with other benchmark methods. The outcomes demonstrate that our approach is superior to others.