High level information fusion through a fuzzy extension to Multi-Entity Bayesian Networks in Vehicular Ad-hoc Networks

This paper presents a novel High-Level Information Fusion architecture based on a fuzzy extension to Multi-Entity Bayesian Networks (MEBN). Modeling both semantic and causal relationships between the existing entities in a specific context, MEBN are deemed a very well-studied and theoretically rich approach that takes advantage of the expressiveness power of First-order Logic, and uncertainty management of Bayesian Networks. However, MEBN lack the capability of modeling the ambiguity which is intrinsic to the knowledge gained through human language. In this paper, a fuzzy extension to MEBN is proposed based on the concept of Fuzzy Bayesian Networks, and a novel ambiguity propagation approach is introduced further. The applicability of the proposed architecture is investigated by implementing a Collision Warning System in Vehicular Ad-hoc Networks. It is shown that our system is capable of not only dealing with both semantic and causal relationships between the existing entities, but it also handles the inherent ambiguity which lies in the input information very efficiently.