Centroid Based Classification Model for Location Distinction in Dynamic Wireless Network

Effective location distinction can help to detect the replication attack towards wireless stations. Instantaneous signal strength information can be used to identify different location information for one certain station. However, most of the previous solutions are under an assumption of a static network. In this paper, we propose a simple centroid based classification model to effectively classify the packets sent by masqueraders among all the packets received based on the aggregate signal strength vectors of packets from multiple access points. The simulation results indicate that the self-location recognition accuracies of our method for static and moving stations achieve 95% and 90%, respectively. Moreover, our method is shown to be very effective in attacker detection, in which attacker locations detection accuracy surpasses 80% even if the attacked targets are moving.