Energy-Efficient UAV-Assisted Computing Offloading and Content Caching for Cloud-Edge Collaborative Networks

Abstract

Cloud-edge collaborative networks, which seamlessly integrate cloud and edge computing capabilities, are a promising paradigm for enhancing network collaboration and performance. In particular, unmanned aerial vehicles (UAVs), functioning as aerial base stations with computing and caching resources, are increasingly used in collaborative network scenarios to offer users flexible services. However, most existing studies focus primarily on either computation-intensive or content-centric tasks, often overlooking the heterogeneous task requirements of applications. These tasks demand that edge nodes provide both computing and caching resources simultaneously to ensure low-latency, immersive user experiences, thereby meeting high standards for quality and interactivity. To address these challenges, we propose an energy-efficient UAV-assisted computing offloading and content caching framework. In this framework, we formulate the joint optimization of the UAV's hovering position, computing offloading, and content caching decisions as an energy consumption minimization problem. Given the nonconvex nature of this problem, we decompose it into two subproblems: one for joint offloading and caching decisions and another for optimizing the hovering position. Furthermore, we develop a deep reinforcement learning (DRL)-based successive convex approximation (SCA) algorithm to achieve a near-optimal solution with low computational complexity. Numerical results demonstrate that the proposed framework effectively utilizes resources in cloud-edge collaborative networks, significantly reducing overall system energy consumption.