Risk-constrained counterfactual freshness routing for reliable status updates in the internet of vehicles

Safety alerts, cooperative perception packets, and vehicle-state updates in the Internet of Vehicles (IoV) are valuable only when delivered before their information becomes stale. This paper proposes Risk-Constrained Counterfactual Freshness Routing (RCFR), which redefines the routing decision object from ranking neighboring nodes by instantaneous or predicted states to evaluating the short-horizon consequence of each candidate forwarding action. For every candidate path, RCFR jointly estimates link survival, end-to-end delay, queue pressure, forwarding progress, task value, and freshness decay, and then selects the action that maximizes task-dependent freshness utility. A conditional edge-diverse backup mechanism further protects safety-critical packets when residual risk remains high. Counterfactual denotes an action-conditioned predictive comparison among alternative forwarding decisions, while risk-constrained refers to a soft routing-score and conditional-backup realization.The risk-constrained mechanism is implemented as a soft routing score and conditional backup rule rather than as a strict probabilistic constraint solver. RCFR is evaluated through complementary SUMO-driven decision-layer screening and Veins/OMNeT++ packet-level simulations in urban and sparse-stress scenarios. In the urban four-roadside unit (RSU) setting, RCFR improves packet delivery ratio (PDR) and deadline satisfaction from 0.9700 to 0.9920, reduces p95 delay by 10.57%, and lowers link failure by 39.50% relative to a reliability-weighted baseline. Packet-level results confirm that the proposed action-conditioned decision mechanism retains delivery and freshness advantages under contention, collisions, queueing, retransmissions, and control traffic. These results demonstrate the effectiveness of action-consequence-aware freshness routing for reliable and timely status-update delivery in dynamic IoV environments.