An evaluation of subjective logic for trust modelling in information fusion

Abstract: Information fusion is to combine information from a variety of sources, or sensors. When the sources are uncertain or contradicting conflict can arise. To deal with such uncertainty and conflict a trust model can be used. The most common ones in information fusion currently is bayesian theory and Dempster-Shafer theory. Bayesian theory does not explicitly handle ignorance, and thus predetermined values has to be hard coded into the system. This is solved in Dempster-Shafer theory by the introduction of ignorance. Even though Dempster-Shafer theory is widely used in information fusion when there is a need for ignorance to be modelled, there has been serious critique presented towards the theory. Thus this work aims at examining another trust models utility in information fusion namely subjective logic. The examination is executed by studying subjective logic using two scenarios from the literature. The results from the scenarios points to subjective logic being a reasonable approach for modelling trust in information fusion.