Multi-sensor configurations in data fusion systems

Abstract

This paper discusses the problems, related to the classification of sensor configurations in multi-sensor fusion systems from the information theory point of view. The criteria, used at present, are defined by the information interactions between the sources. They are qualitative in nature and allow for subjective interpretation, which significantly complicates the classification process, based on them. The present work proposes new criteria and a new type of classification, defined within the framework of the partial information decomposition - a recent extension of the information theory. The proposed criteria, unlike the existing ones, are quantitative, ensure objectivity in classification and allow for direct comparison and evaluation of different sensor configurations.