Anomaly Detection Based on System Call Classification

The aim of this paper is to create a new anomaly detection model based on rules. A detailed classification of the LINUX system calls according to their function and level of threat is presented. The detection model only aims at critical calls (i.e. the threat level 1 calls). In the learning process, the detection model dynamically processes every critical call, but does not use data mining or statistics from static data. Therefore, the increment learning could be implemented. Based on some simple predefined rules and refining, the number of rules in the rule database could be reduced dramatically, so that the rule match time can be reduced effectively during detection processing. The experimental results clearly demonstrate that the detection model can effectively detect R2L, R2R and L2R attacks. Moreover the detected anomaly is limited in the corresponding requests, but not in the entire trace. The detection model is fit for the privileged processes, especially for those based on request-responses.