Vehicle Edge Computing Network Service Migration Strategy Based on Multi-agent Reinforcement Learning

In order to address the decision-making problem in vehicular edge computing networks, this paper presented a vehicle edge computing service migration strategy based on multi-agent deep learning. This strategy employs the multi-agent deep deterministic policy gradient algorithm, enabling vehicles to learn from incomplete system information and make distributed online migration decisions with only partial observations. Users collaborate and compete to achieve common goals, making the system more flexible and enhancing the overall benefits and stability. Simulation results using real datasets demonstrate that the proposed strategy converges rapidly and exhibits superior performance, robustness, and stability in scenarios with varying numbers of users and task arrival rates compared to other benchmark strategies.