Semantic Intruder Detection System in WSN

Today, surveillance systems play a key role in highly secured areas. The most common surveillance system, CCTV, produces a continuous image and has various problems such as high cost, need for continuous human intervention, and an inherent difficulty in locating and identifying the intruder due to the presence of obstacles. This paper proposes a scheme to classify a person´s movement as intrusive or non-intrusive in a semantic way. The classification is aided with the help of sensor network. The sensor network is considered to be heterogeneous. It uses a PIR sensor for movement detection, an RFID sensor for identifying the person and an image sensor for verification. This system uses both wired and wireless networks for efficient data transfer. Data from either networks are synchronised using time and location information. The proposed Intruder Detection System identifies a person as an intruder, if there is any mismatch in the sensed information.