Dempster–Shafer information filtering framework: Temporal and spatio-temporal evidence filtering

Abstract

This paper presents an information processing framework for distributed sensor networks. The framework is capable of directly processing temporally and spatially distributed multimodality sensor data to extract information buried in the noise clutter. Moreover, we introduce distributed algorithms to implement spatio-temporal filtering applications in grid sensor networks within the context of the framework. The proposed framework is based on the belief notions in Dempster-Shafer (DS) evidence theory and evidence filtering method. Further analysis is done by exploiting a fire propagation scenario when high noise is present in the sensed data. We compare intuitively appealing results against DS fusion method to grant further credence to the proposed framework.