Trust Model for CAV Networks Cuts Misinformation Impact by 91%

Connected and autonomous vehicles (CAVs) increasingly rely on digital route guidance for traffic management, but when that guidance is unreliable or adversarial, the system must account not only for traffic flow adaptation but also for the evolution of user trust in the information source. A new paper from researchers Eunhan Ka and Satish V. Ukkusuri introduces a coupled day-to-day traffic assignment and trust-evolution framework that models how trust—modeled as an aggregate behavioral reliance state via a Beta evidence model—responds to repeated guidance errors. The within-day congestion uses Lighthill-Whitham-Richards network loading, while day-to-day route choice follows bounded-rationality logit learning with trust-dependent reliance on external guidance. The theoretical analysis establishes stationary equilibria, a conservative stability guide, a weighted compliance index for population-level vulnerability, and an asymmetric recovery law explaining post-attack trust hysteresis.

Numerical experiments on the Sioux Falls network (with an Anaheim robustness check) reveal a critical threshold-based resilience mechanism. Below the trust-activation threshold, the attack remains behaviorally stealthy, and dynamic trust provides almost no attenuation. Above the threshold, trust erosion reduces the impact of a fixed-trust attack by approximately 91% in Sioux Falls and 85% in Anaheim. Interestingly, higher CAV penetration increases fixed-trust vulnerability while preserving dynamic attenuation. Perhaps most alarmingly, traffic performance can recover before trust does, creating a 77-day hidden vulnerability window during which the network appears normal but trust remains eroded. These findings provide a trust-aware modeling basis for designing resilient CAV-enabled traffic networks that can withstand misinformation attacks.