Dempster–Shafer evidence theory based IFA detection approach towards mixed attacks in VNDN

Vehicular Named Data Networking (VNDN) enables communication based on content names rather than vehicle addresses. This approach effectively mitigates the limitations of traditional vehicular networks that rely on TCP/IP-based communication. However, due to various network attacks, VNDN faces significant cybersecurity risks, which severely impact network performance and efficiency. To address these issues, this paper proposes a mixed attacks detection method based on the Dempster–Shafer evidence theory, integrating the Particle Swarm Optimization algorithm (DS-PSO). The method first extracts three key feature indicators: the information entropy offset of the Interest packet names, the cache offset in the content store, and the difference between the number of Interest packets sent and Data packets received by routing nodes per unit time. These indicators are normalized and used as evidence in DS evidence theory. The basic probability assignment of this evidence is then transformed into a parameter selection problem, and PSO is employed to optimize this selection by finding the optimal solution. Building on this, the DS evidence theory is used to fuse the obtained evidence, and the overall network security state is determined based on the fusion results, identifying and detecting existing network attacks. Our experimental results demonstrate that, compared to other methods, the proposed detection method effectively improves detection accuracy and reduces error rate not only in the various single-attack scenarios but also in the mixed-attack scenario.