Semantic intrusion detection with multisensor data fusion using complex event processing

Abstract

Complex Event Processing (CEP) is an emerging technology for processing and identifying patterns of interest from multiple streams of events in real/near real time. Sensor network-based security and surveillance is a topic of recent research where events generated from distributed sensors at an unpredictable rate need to be analysed for possible threats and respond in a timely manner. Traditional software architectures like client/server architecture where the interactions are pull-based (DBMS) do not target the efficient processing of streams of events in real time. CEP which is a push-based system can process streaming data to identify the intrusion patterns in near real time and respond to the threats. An Intrusion Detection System (IDS) based on single sensor may fail to give accurate identification of intrusion. Hence there is a need for multisensor based IDS. A multisensor-based IDS enables identification of the intrusion patterns semantically by correlating the events and context information provided by multiple sensors. JDL multisource data fusion model is a well-known research model first established by the Joint Directorate Laboratories. This paper proposes JDL fusion framework-based CEP for semantic intrusion detection. The events generated from heterogeneous sensors are collected, aggregated using logical and spatiotemporal relations to form complex events which model the intrusion patterns. The proposed system is implemented and the results show that the proposed system out performs the pull-based solutions in terms of detection accuracy and detection time.