An architecture for wireless intrusion detection systems using artificial neural networks

Abstract

The majority existing wireless intrusion detection systems identifies intrusive behaviors are based on the exploration of known vulnerabilities called signatures of attacks. With this mechanism, only known vulnerabilities are detected which leads to bringing the necessity of new techniques to add in the system. This work considers an architecture for intrusion detection in wireless network based on anomaly. The system is capable to adapt itself to a profile of a new community of users, as well as recognizing attackswith different characteristics than those already known by the system, by considering changes from normal behavior. The system uses artificial neural networks in the processes of detecting intrusions and taking countermeasures. A prototype is implemented and submitted to some simulations and tests, with three different types of attacks of Denial of Service (DoS). © 2010 Springer Science+Business Media B.V.