Analyzing the degree of conflict between bodies of evidence based on a new distance in data fusion

Abstract

Dempster’s combination rule may produce some unreasonable results when dealing with a combination of the conflicting evidence in evidence theory of Dempster–Shafer. Therefore, analyzing the degree of conflict between the bodies of evidence is essential to evaluate the applicability of Dempster’s rule. A new probability function, which is called a supporting probability function, is proposed to describe the correlation between evidences, and its distance is proposed to measure the distance between bodies of evidence. Combining this distance with classical conflict coefficient, a new method of evaluating the applicability of classical Dempster’s combination rule is presented. A weighted average approach to combine the conflicting evidences based on a supporting probability distance between the bodies of evidence is proposed. Numerical examples are given to illustrate the interest of the proposed approach.