Deep Reinforcement Learning-Based Task Offloading for Multi-User Distributed Edge Computing
Zhengtian Shi, Zhisheng Zhang, Min Dai, Zhijie Xia, Haiying Wen, Fan Huang · 2024
Edge Computing (EC) deploys multidimensional resources, including computation, storage, and communication, at the network edge to fulfill certain applications with stringent latency requirements. In the edge environment, computing tasks can be offloaded to mobile edge computing servers (MECS) for reduced task processing latency. However, traditional methods of task offloading in edge computing are inadequate for complex environments demanding high real-time performance. This paper focuses on studying the computational task processing problem in a multi-user distributed edge environment and propose a deep reinforcement learning-based multi-user distributed edge computing task offloading method aimed at minimizing task processing latency. The proposed method effectively addresses the computational task processing challenge in dynamic and complex edge environments with high real-time demands. Simulation results show that the proposed method has superior performance in terms of both task processing delay and algorithm execution time.