Empirical analysis of generalised uncertainty measures with dempster shafer fusion

Abstract

Data Fusion within the evidential reasoning framework is a well established, robust and conservative technique to fuse uncertain information from multiple sensors. A number of fusion methods within this formalism were introduced including Dempster-Shafer Theory (DST) Fusion, Dezert Samarandche Fusion (DSmT), and Smets' Transferable Belief Model (TBM) based fusion. However, the impact of fusion on the level of uncertainty within these techniques has not been studied in detail. While the use of Shannon entropy within the Bayesian fusion is well understood, the measures of uncertainty within the Dempster-Shafer formalism is not widely regarded. In this paper, an uncertainty based technique is proposed to quantify the evolution of DST fusion. This technique is then utilised to determine the optimal combination of sensor information to achieve the least uncertainty in the context of an aircraft identification problem using Electronics Support Measure (ESM) sensors.

Sensor fusion in smart camera networks for Ambient Intelligence

“…K is used to highlight the conflict of the two sources of evidence m i and m j by aggregating the total contradiction, or reversely the agreement as 1 − K [84].…”

“…A similar, but empirical, study into sensor uncertainty is performed in [84] using D-S theory as the fusion method. They propose an uncertainty based technique for quantifying the evolution of D-S theory based fusion.…”

Experimental study of uncertainty measures with sensor fusion techniques

“…INTRODUCTION Information fusion commonly deals with the integration of uncertain information from multiple sensors. Information often contains uncertainties, which are usually related to physical constraints, detection algorithms, and the transmitting channel of the sensors [1].…”

“…An informed decision maker often relies on various forms of data fusion models to assist him/her to assess the current situations, uncertain information often exists on all levels of fusion process [1].…”

Uncertainty evaluation for an ultrasonic data fusion based target differentiation problem using Generalized Aggregated Uncertainty measure 2

“…On the other hand, uncertain information often exists on all levels of fusion process which are usually related to physical constraints, detection algorithms, and the transmitting channel of the sensors [11]. Therefore, it is important to have an uncertainty evaluation after sensor fusion for better decision making.…”