Physical risk aware agents improve communication reliability in vehicle networks
Physical-Risk-Aware Decentralized Agents for Constrained Vehicle Communications
Multiagent Systems
Summary
Vehicles in a network need to talk to each other, but each vehicle only knows what it can see nearby. This limited view can make communication unreliable across the whole network. The authors created a system where each vehicle uses local information and a shared policy learned from training to pick communication resources carefully, prioritizing safety-related messages. Their method, called PA-STGAT, helps maintain reliable communication even in tricky traffic situations, balancing safety, network use, and message delivery. Tests show it works better than some other approaches but has tradeoffs in how often repairs to the network are needed.
What this means in practice
- •For automotive communication engineers: Implement decentralized V2X communication controls that prioritize safety and reliability using local vehicle data and learned policies.
- •For network protocol developers: Design resource allocation protocols for vehicle networks that use risk-aware coordination signals to improve message delivery under mobility.
Authors
Haoran Peng, William Weijia Zheng, Hang Liu, Ying-Jun Angela Zhang
Abstract
Decentralized V2X resource control is challenging because each vehicle has only local observations, while reliability depends on network-wide interactions. We develop a physical-risk-aware spatio-temporal graph attention (PA-STGAT) framework for agentic V2X control. Each vehicle perceives its local interaction graph, retains short-term temporal context, and independently selects its PC5 resource and power. Network outcomes are used during centralized training to refine the shared policy. Transmitter-specific credits map reliability failures to individual vehicles. A dual-priced graph term further converts network violations into coordination signals for discrete resource selection, with higher priority on short time-to-collision links. Experiments show higher raw-policy reliability than the tested graph-based baselines. Under filtered execution, PA-STGAT meets the nominal empirical communication requirements across all five seeds, with tradeoffs among utility, delivery reliability, resource occupancy, and filter-induced repair. Compared with GNN+MATD3, it requires fewer resource repairs but achieves lower utility and mean delivery reliability. Mechanism ablations show that removing dual-priced graph coordination causes resource-selection collapse. Density and mobility tests characterize the operating limits of empirical constraint satisfaction.