Abstract:

A network selfprotection mechanism based on abnormality analysis is presented aiming at deficiencies of current network security protection technologies. To improve the ability of detection and protection against network attacks, a network attack detection algorithm based on the multivariate abnormality analysis is proposed. The algorithm uses a metric of abnormal distance to classify the network flow into different types and prioritize the routing of the different network flow packets. In this way, the normal traffic flows can suffer the least impact from network attacks. Experiments show that our mechanism can significantly improve the network system’s protection ability against distributed denial of service attacks.

Key words: selfprotection, network attack, abnormality analysis, mechanism