A Novel Multi-Criteria Discounting Combination Approach for Multi-Sensor Fusion

Abstract

Belief function theory manages uncertain information and offers useful combination rules for multi-sensor fusion. However, when sensor readings are in conflict or even unreliable, the quality of the fusion result is significantly affected. Recently, many discounting approaches have been proposed to combine unreliable sensor readings. The discounting factors involved in these methods are often determined based on a single criterion which is not sufficient in general to obtain a precise assessment of the reliability degrees of the sources to combine. In this work, that is why we propose a novel discounting combination approach, in which the reliability factors are obtained by using the multi-criteria strategy. Our discounting combination method includes two main steps. The first step to assess the sensor’s reliability is based on belief function-based technique for order preference by similarity to ideal solution (BF-TOPSIS). The second step is to discount and global combine all involved sensor readings according to their degree of reliability with proportional conflict redistribution-6 (PCR6) rule. Several simulations and comprehensive comparisons with classical approaches are given to show the efficiency of our proposed method.