Mobility-Aware Dynamic Service Migration in Communication and Computing Integrated VNETs

The Communication and computing integrated vehicular networks have emerged as a reliable platform for intelligent vehicular applications. These networks leverage mobile edge computing technology to enable computing capabilities at the network edge closer to vehicles. However, the mobility of vehicles could break service continuity and make it difficult to accomplish intelligent services within tolerant latency under communication and computing resource-constrained conditions. To overcome these challenges, this paper proposes a mobility-aware service migration approach to effectively reduce service latency under the expected migration cost constraint. The key aspect of this approach is leveraging the mobility information of vehicles to search for the best migration strategy. This paper utilizes the wireless sensing technology of the base stations to acquire real-time mobility information of vehicles. The mobility information and channel states are jointly considered to make decisions on service migration as well as resource allocation strategies via a decentralized multi-agent proximal policy optimization (MAPPO) algorithm. Simulation results demonstrate that the proposed mobility-aware service migration approach can effectively reduce the average service latency compared to other benchmark approaches.