Substantiating Security Threats Using Group Outlier Detection Techniques

Abstract—With the increasing dependence on wireless LANs (WLANs), businesses, educational institutions and other organi-zations are in need of a reliable security mechanism. The latest security protocol, the IEEE 802.11i assures rigid security for WLANs with the support of IEEE 802.1x protocol for authen-tication, authorization and key distribution. Nevertheless, fresh security threats are emerging often to oust these new defense mechanisms. Further, many organizations based on superficial vendor literature, believe their wireless security is sufficient enough to prevent any unauthorized access. Having wide ranging options for security configurations, users are camouflaged into profound uncertainty. This volatile state of affairs has prevented many organizations from fully deploying WLANs for their secure communication needs, though WLANs may be cost effective and flexible. In this paper, we present an anomaly based mechanism to detect and substantiate both known and unknown security threats in WLANs. Our method exploits both timing and behavioral anomalies. We first observe for timing and/or behavior anomalies during the security association process and use outlier based data association approaches to substantiate their legitimacy. The proposed concept was tested on our experimental setup and the results obtained from EAP TLS authenticated hosts are presented here. I