An Adaptive Belief Representation for Target Tracking using Disparate Sensors in Wireless Sensor Networks

Sensor diversity has been shown to increase the accuracy and robustness of Wireless Sensor Network (WSN) target tracking. However, difficulties can arise due to disparity between sensor types. This paper seeks to address specifically those sensor measurements which require additional information from another source before they can be used to create a location estimate; such as a microphone mote, which requires knowledge of the target's acoustic power before it can estimate a distance. A novel representation, called the Adaptive GRiD (Grid Representation of belief Distribution), is presented for such sensor measurements which facilitates fusion with other measurement types, overcoming disparity. This is accomplished using a state space which can expand to track extra parameters of the target apart from location, and subsequently contract when those parameters are no longer necessary. In this way the tracker can adapt to a variety of different sensor types whose measurements are mathematically related to target properties besides location. The proposed representation is evaluated for its effectiveness and suitability and shows promising results.