Sensor collaboration in event detection

In this paper we introduce a probabilistic detection model for sensors. A sensor can detect a close event based on sensitivity measures. This sensitivity is a function of the distance between the sensor and the event and also includes thermal noise. This leads to a probabilistic detection of events. We then describe a Local Fusion Detection algorithm which gathers the measurements of neighboring sensors to build a decision related to events that may occur in a given area. We compare the Local Fusion Detection (LFD) technique with the usual Single Node Detection (SND) technique where each sensor decides whether or not to report the detection of an event based only on its local information. We build an analytical model to compare both techniques. We show that LFD outperforms SND. This remains true when energy consumtion is taken into account and whether the nodes are located on a regular grid or randomly distributed.