Distributed RSSI processing for intrusion detection in indoor environments

Abstract

In the context of WSNs, the RSSI has been traditionally exploited for localization, distance estimation, and link quality assessment. Recent research has shown that variations of the signal strength in indoor environments where nodes have been deployed can reveal movements of persons. Moreover, the time-histories of the RSSI over multiple links allow reconstructing the paths followed by the persons inside the monitored area. This approach, though effective, requires the transmission of multiple entire RSSI time-histories to a sink where these signals are processed, increasing latency and power consumption. This work aims at applying distributed processing of the RSSI signals for intrusion detection. Through distributed processing, the nodes are able to detect and localize moving persons autonomously. The latency and power consumption of the proposed intrusion detection system is minimized by transmitting to the sink only alert notifications related to significant events. Moreover, an accurate time-synchronization allows the nodes to keep the radio off most of the time. The proposed system was able to detect the intrusion of a person walking inside the monitored area, and to correctly keep track of the path he had followed. Possible applications of such a system include surveillance of critical buildings, support to emergency workers in locating people e.g. during fires and earthquakes, and to police in hostage situations or terrorist attacks.