Communication-Aware Distributed Control for Cooperative Ramp Merging in Vehicular Networks

Abstract

Cooperative ramp merging requires reliable coordination among connected vehicles, but practical NR V2X Mode 2 communications may suffer from delay, packet loss, and resource collision due to distributed sidelink resource selection. Existing distributed control methods often assume ideal communication or treat communication impairments as external disturbances, without explicitly linking communication states to vehicle safety constraints. Accordingly, this paper proposes a communication aware distributed model predictive control (CA-DMPC) frame work for cooperative ramp merging. A Markov-chain-based model is developed to characterize NR V2X Mode 2 resource reservation and re-evaluation, including collision persistence, resource reselection, and re-evaluation failure. Predicted communication states are converted into spatial prediction uncertainty and incorporated into MPC safety constraints through an adaptive safety margin. To improve real-time performance, the communication-control optimization problem is decomposed into tractable subproblems, and the vehicle-control component is reformulated as a quadratic programming problem. Simulation results show that the proposed framework achieves smoother control, more stable inter-vehicle spacing, better trajectory tracking, and lower computational cost than benchmark methods.