A Lightweight Intrusion Detection System for In-Vehicle Bus Networks

Vehicle networks designed based on Service-Oriented Architecture (SOA) present considerable benefits while they are subject to substantial cybersecurity vulnerabilities. For the Controller Area Network (CAN) protocol with weak security protections, a lightweight Intrusion Detection System (IDS) based on Renyi entropy is proposed. The proposed scheme employs the real-vehicle dataset for offline training of the IDS, which is subsequently utilized for real-time online detection. By employing the more sensitive Renyi entropy in place of the traditional Shannon entropy, the detection accuracy has been significantly improved. Comparative studies with other methods, along with adaptability studies across three vehicle models demonstrate that the proposed scheme exhibits superior detection performance and adaptability. The average detection accuracy exceeds 98% across all scenarios.