Multi-sensor distributed detection in presence of communication cost

Abstract

The subject of multi-sensor data fusion (or 'sensor fusion' in short) has grown explosively and a multitude of approaches and areas have been covered over the past decade ever since the first related paper was presented by Tenney and Sandell in 1981. Most citations on multi-sensor distributed detection made an assumption that transmission of information from the sensor to the fusion center as well as to each other sensors is done at no cost. The aim of this dissertation is to present a taxonomy of multi-sensor data fusion as well as a detection fusion architecture for addressing multi-sensor distributed detection problems; and to perform a comparative analysis on the optimum detection approaches. These approaches are based upon minimization of the cost functionals associated with the probability of false alarm and the probability of miss taken into consideration the communication cost given that a certain amount of risk can be tolerated. A simple detection fusion model is also presented to clarify the issues further.