RECOGNITION AND PREPROCESSING OF INTRUSION DATA IN WIRELESS SENSOR NETWORK?

In wireless sensor networks (WSNs), the significant deviation of measurements from the normal pattern of sensed data are considered as intrusion. The main sources of intrusion are noise and errors, events, and malicious attacks on the network. In this work, we focused on the problem of detecting the intrusion by proposing a model that is based on distribution of sensor data stream approximately over the data space. With the set of data collected from Intel Berkley lab, we processed this data by our proposed scheme and evaluated it. We find that experimental evaluation of our proposed scheme can achieve high precision rating for detecting intrusion.