FedTO: Mobile-Aware Task Offloading in Multi-Base Station Collaborative MEC

With the proliferation of the Internet of Things (IoT), mobile edge computing (MEC) has great potential to achieve low latency, high reliability, and low energy consumption. However, in collaborative MEC environments, user movement and task migration may cause task transmission and processing delays, resulting in elevated task response times. Therefore, system performance and user experience need to be ensured by rational task offloading and resource management. At the same time, the protection of user data privacy is becoming increasingly important as a challenge to be overcome. To address the problems of intense resource competition and privacy leakage in MEC, the fed erated learning for the T D3-based task o ffloading (FedTO) algorithm is proposed. The algorithm has a dual objective of energy consumption and task response time while protecting user privacy. It employs a cryptographic local model update and aggregation mechanism and uses deep reinforcement learning (DRL) to obtain an efficient task offloading decision. Based on the mobile trajectories of real devices, and the pre-deployment of base station locations, experimental results show that the FedTO algorithm ensures task data security. It also effectively reduces the total energy consumption and average task response time of the system, which further improves the system utility.