A Novel Approach to Asynchronous , Decentralized Decision - Makingin

In a radical shift from the previous doctrine of repower, attrition, and destruction during the Vietnam era, the current \airland battle" military doctrine is characterized by four key ingredients { depth, initiative, agility, and synchronization. The term, synchronization, implies bringing to bear, at one time and place, the combined power of maneuver, artillery, air, deception, and other systems so as to strike the enemy again and again with an impact that exceeds the sum of all the parts. Throughout military history, in general, synchronization has been achieved through centralized command, control, and communication (C 3), i.e. one where all intelligence is collected and decisions generated at a centralized, sequentially executing entity. While the military synchronization is laid down by a higher command and must be obeyed, the lower level units are granted military initiative as to how to achieve the requirements of synchronization. However, at any given level in the military hierarchy, the synchronization requirement is logically inconsistent with the other three directives, particularly initiative. Evidence from World War II and Operations \Just Cause" and \Desert Storm" indicate that excessive synchronization, realized through centralized C 3 , may have caused (i) unnecessary mission failure, (ii) fratricide, and (iii) failure to take advantage of the maximal distributed and concurrent intelligence from the diierent combat units. This paper proposes a novel asynchronous, decentralized decision-making algorithm for military command and control. Unlike centralized command and control, in the proposed approach, all entities { infantry, tanks, and aircrafts, of the system are granted signiicant autonomy, subject to strict guidelines and directives, and provided with the necessary computing power, intelligence gathering facilities, communications gear, and decision-making aids. The overall decision task is distributed among these entities. As a result, every entity's execution is asynchronous, concurrent, and autonomous , leading to eecient and fast decision making. Asynchronous, distributed, discrete-event simulation techniques are utilized to model the problem of military command and control. This ability to model is made possible by the fact that the transfer of discrete data, geographically-dispersed decision-making entities, asynchronously generated stimuli, and the presence of feedback in the data transfer process are the properties of asynchronous, distributed discrete-event simulation algorithm. This paper also presents an implementation of the proposed approach on a testbed of 40+ SUN sparc workstations, conngured as a loosely-coupled parallel processor. The goal is to evaluate the performance of the decentralized C 3 algorithm through simulation, under representative battleeeld scenarios, …