MARL-Based Joint Optimization of Service Migration and Resource Allocation in MEC

Ziqi Lin, Zhenli He, Xiaolong Zhai, Yuanfei Xiao

Abstract: Mobile Edge Computing (MEC) extends computational resources to the network edge, enabling low-latency services for users and IoT devices. While existing research advances service migration, it often isolates migration from resource allocation, leading to suboptimal edge server utilization and increased response delays. Moreover, the impact of migrating user-specific service contexts within network topologies is often overlooked, which undermines the effectiveness of migration strategies and increases costs. To address these issues, we propose a joint optimization of service migration and resource allocation in MEC, modeled as a Multi-Agent Markov Decision Process (MAMDP). We propose an algorithm based on Multi-Agent Reinforcement Learning (MARL). By employing Multi-Agent Proximal Policy Optimization (MAPPO) and Karush-Kuhn-Tucker (KKT) conditions, our approach optimizes both service migration and resource allocation strategies, thereby enhancing service quality. Simulations demonstrate that our method significantly outperforms benchmarks, achieving substantial reductions in response delays, service failures, and migration costs.