Evidence updating in a heterogeneous sensor environment

Abstract

Nodes in a distributed sensor network (DSN) gather and fuse information generated by heterogeneous sources to arrive at local decisions targeted at achieving a given global mission objective. These sources typically possess very diverse scopes of 'expertise' or frames of discernment (FoDs) that renders updating a knowledge base from the evidence received a challenging task. In this paper, we present a Dempster-Shafer (DS) theory based evidence updating strategy that accommodates such non-identical FoDs. It is composed of a linear combination of the available evidence and incoming evidence conditioned to the source it is being generated from. The linear combination weights can be used to accommodate differences in source reliability and 'inertia' of the existing knowledge base. Strategies to choose these weights are also proposed.