International Conference on Mechanical Engineering and Technology (ICMET-London 2011)
183 Host Based Intrusion Detection System
Download citation file:
- Ris (Zotero)
- Reference Manager
Host based intrusion detection has emerged as an important approach to safeguard the system from unauthorized users. It detects possible attacks on a system, on which the intrusion detection system runs. In this paper we have proposed a model capable of distinguishing an authorized user from an unauthorized one by studying the pattern of usage of the system, even after compromising the basic authentication protocol i.e. username and password. Our model of intrusion detection uses different aspects of data mining and probability theory. The proposed model deals with the analysis of system log files; data mining concepts are applied to extract the behavior pattern of the user and probability theory is used to predict the authorized user. The average success rate is above 94% for a system which has rich history files.