Distributed Collaborative Anomaly Detection for Trusted Digital Twin Vehicular Edge Networks

The vehicular networks are vulnerable to cyber security attacks due to the vehicles’ large attack surface. Anomaly detection is an effective means to deal with this kind of attack. Due to the vehicle’s limited computation resources, the vehicular edge network (VEN) has been proposed provide additional computing power while meeting the demand of low latency. However, the time-space limitation of edge computing prevents the vehicle data from being fully utilized. To solve this problem, a digital twin vehicular edge networks (DITVEN) is proposed. The distributed trust evaluation is established based on the trust chain transitivity and aggregation for edge computing units and digital twins to ensure the credibility of digital twins. The local reachability density and outlier factor are introduced for the time awareness anomaly detection. The curl and divergence based elements are utilized to achieve the space awareness anomaly detection. The mutual trust evaluation and anomaly detection is implemented for performance analysis, which indicates that the proposed scheme is suitable for digital twin vehicular applications.