paper

Uncertainty Representation and Evaluation for Modelling and Decision-making in Information Fusion

Volume Number:
13
Issue Number:
2
Pages:
Starting page
198
Ending page
215
Publication Date:
Publication Date
1 December 2018
Author(s)
J. P. De Villiers, G. Pavlin, A.-L. Jousselme, S. Maskell, A. De Waal, K. Laskey, E. Blasch, P. Costa

paper Menu

Abstract

In this paper, the uncertainties that enter through the life-cycle of an information fusion system are exhaustively and explicitly considered and defined. Addressing the factors that influence a fusion system is an essential step required before uncertainty representation and reasoning processes within a fusion system can be evaluated according to the Uncertainty Representation and Reasoning Evaluation Framework (URREF) ontology.

The life cycle of a fusion system consists primarily of two stages, namely inception and design, as well as routine operation and assessment. During the inception and design stage, the primary flow is that of abstraction, through modelling and representation of real-world phenomena. This stage is mainly characterised by epistemic uncertainty.

During the routine operation and assessment stage, aleatory uncertainty combines with epistemic uncertainty from the design phase as well as uncertainty about the effect of actions on the mission in a feedback loop (another form of epistemic uncertainty). Explicit and accurate internal modelling of these uncertainties, and the evaluation of how these uncertainties are represented and reasoned about in the fusion system using the URREF ontology, are the main contributions of this paper for the information fusion community. This paper is an extension of previous works by the authors, where all uncertainties pertaining to the complete fusion life cycle are now jointly and comprehensively considered. Also, uncertainties pertaining to the decision process are further detailed.