A study of Intrusion Discovery Method Using Genetic Neural Network

COMPUTER SCIENCE

An Intrusion detection method is broadly classified as Anomaly based and Rule based detection methods. Anomaly based systems look for strange system behavior by observing the deviation from a baseline of normal behavior. Hence the anomaly based system has to be trained for the normal behavior. The training results in an ‘activity profile’ which represents the normal usage for a particular user over definite period of time. This acts as the baseline for the anomaly based IDS and any event that deviates from this baseline is reported as anomalous. Statistics based anomaly detection is best suited for Wireless Adhoc Networks. On other hand rule based IDS looks for a malicious event based on the rule set that is already available and customized. Wireless ad-hoc networks are increasingly being used in the tactical battlefield, emergency search and rescue missions, as well as civilian ad-hoc situations like conferences and classrooms due to the ease and speed in setting up such networks. As wireless ad-hoc networks have different characteristics from a wired network, the intrusion detection techniques used for wired networks may no longer be sufficient and effective when adapted directly to a wireless ad-hoc network. Existing methods of intrusion detection have to be modified and new methods have to be defined in order for intrusion detection to work effectively in this new network architecture. In this paper, we will first provide an introduction to wireless ad-hoc networks and thereafter an introduction to intrusion detection. We will then present various existing intrusion detection techniques that can be adapted to wireless ad-hoc networks and finally propose a hybrid intrusion detection system for wireless ad-hoc networks.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.