Feature-based multisensor fusion and performance quantification using sufficient statistics

Synergistic use of multiple-sensor data for improved situational awareness has been of great interest in distributed signal processing. The key factor in multisensor fusion (MSF) deals with accountability and information extraction. Accountability refers to the level of performance improvement attributable to each sensor. That is, in order for us to make a judicious decision on the final sensor configuration, performance ambiguity must be minimized. Furthermore, MSF algorithms must be able to extract all the useful information from raw sensor data. In short, we need to develop a meaningful metric that measures the extent to which the MSF performance approaches the theoretical upper limit similar in concept to the Cramer-Rao bounds (CRB) in parameter estimation. In this paper, we develop an interesting MSF algorithm and introduce a meaningful metric to quantify the degree of information loss associated with each sensor as well as the entire suite of sensors. We illustrate the utility of our algorithm with one insightful example