An Edge-Assisted Internet-of-Vehicles Computing Framework for Fair Tail-Risk Allocation in Cooperative Autonomous Driving

Connected automated driving increasingly relies on cooperative perception from onboard sensors, roadside units (RSUs), smart traffic lights, and vehicle-to-everything (V2X) links, but communication uncertainty can concentrate residual risk on vulnerable road users (VRUs). This study proposes an Ethical-Improved risk-allocation objective for edge-assisted Internet of Vehicles (IoV) cooperative autonomous driving. The objective internalizes responsibility as a bounded risk weight, normalizes equality and maximin terms, and adds explicit VRU tail-risk and VRU/Ego ratio penalties. The evaluation is organized into two strictly separated tracks. In the planner-objective track, Ethical-Improved reduces physical collision rate, aggregate harm, inequality, and VRU tail risk relative to Standard, Selfish, and Ethical-Orig baselines. In the communication-aware IoV track, edge-assisted cooperative perception achieves the best within-track collision-proxy and normalized-harm performance under sampled packet delivery ratio (PDR), age of information (AoI), RSU coverage, perception confidence, and edge delay conditions. Physical collision rate and communication-conditioned collision proxy are reported separately and are not numerically compared across tracks. The results support auditable tail-risk regularization as a controlled-simulator design principle for fairer cooperative autonomous driving, while external validation remains necessary before deployment-oriented claims can be made.