Towards Mitigating Communication Latency Influence in Connected Vehicle Networks by Stochastic Decentralized Model Predictive Control

Communication delays in Connected and Automated Vehicle (CAV) networks significantly impact decentralized optimization-based coordination, increasing risks of collisions and degrading system performance. Existing methods are limited by real-time computational challenges, vulnerability to outdated data, scalability constraints, and difficulties in managing uncertainties. This paper presents a Stochastic Decentralized Model Predictive Control (SDMPC) framework to mitigate the adverse effects of communication delays by incorporating a novel stochastic approximation method for modeling uncertainties. Our approach provides a tight probabilistic bound on safety constraints, ensuring accurate trajectory predictions and improved coordination. Simulation results show that the proposed SDMPC framework reduces the average trajectory deviation and lowers collision risks compared to conventional methods under various communication latency conditions. These improvements make SDMPC an effective solution for large-scale CAV networks, enhancing both safety and efficiency.