MEBN Logic: A Key Enabler for Network Centric Warfare

Abstract

Among the lessons learned from recent conflicts stands the dramatic change in the very way wars are fought. There are no more clear-cut enemies or allies; rules of engagement have become increasingly fuzzy; guerrilla and insurgent tactics are now commonplace: in short, the battlespace is a very different place from what it used to be. Furthermore, advances in sensor technology and network computing have brought a new element to the complex equation of warfare: information overload. Nowadays, instead of merely gathering information and displaying assets, command and control systems must be able to fill the gap between the glut of information arriving from a networked grid of sensors and the capacity of human commanders to make sense of it. In short, the quest today is for systems that work under the knowledge paradigm. Systems must automatically provide decision makers with a clear picture of what is happening, how it relates to the current situation, and what are the options and their respective consequences. Facing this challenge with technologies of the past is a recipe for failure. New, more powerful approaches are needed. The objective of this paper is to argue for two claims: (1) Bayesian decision theory is an appropriate technology for modeling human decision-making in complex, ambiguous scenarios; and (2) Bayesian reasoning technology is a promising enabler for Network Centric Warfare. To support both claims, we have applied Multi-Entity Bayesian Networks (MEBN) to model a historical tactical decision from the naval domain. MEBN is a breakthrough Bayesian reasoning system in which complex probabilistic models are constructed from modular components that can be replicated and combined in an infinite variety of ways. MEBN allows models to capture important and subtle aspects of objects and their interrelationships that would be impossible to model using existing technologies. We provide a brief overview of modeling in MEBN and then present our model and the outcome of applying it to a historical scenario. Our results clearly support the validity of our approach.