A Novel Method of Multi-sensor Information Fusion Based on Comprehensive Conflict Measurement

Abstract

AbstractThe Internet of things expands the ability of human beings to perceive the surrounding environment, bringing a great challenge to the multi-sensor data processing. Evidence theory, one of the most effective processing technologies, is commonly employed in the multi-sensor information fusion. However, many counter-intuitive results of multi-sensor data fusion may be obtained when fused evidence is highly conflicting. In this study, a new comprehensive method for calculating the entropy of each evidence is proposed, with the goal of improving information volume measurement. In addition, a conflict measure method of multi-sensor evidence is introduced, which can calculate the weighted average evidence, by synthesizing vector space and evidence distribution. Finally, the pre-processed body of evidences have been merged based on the evidence theory. The proposed multi-sensor fusion approach based on comprehensive conflict measurement produces a more credible fusion outcome compared to other approaches, according to experimental results.KeywordsEvidence theoryConflictEntropyVector space