Skip to Main Content
In this paper we deal with the design and analysis of an intelligent multi-agent-based architecture for synchronized real-time situational understanding, awareness, decision-making, and control in a geographically networked distributed computing environment. In particular, we focus here on the design and implementation of a middleware framework for agent intra and inter-node communication as well as computing nodes synchronization. While the proposed work finds applications in many areas including networked chemical sensors, large key infrastructures and resources such as highways and transportations, here as application of the proposed method we consider the challenging scenario case of a set of distributed collaborating radars (multi-agent) system geographically distributed over a large terrain environment with several moving targets. Here, each radar agent, taken separately, does not have the capabilities and resources to span the monitoring of the totality of a given large terrain. However, when collaborating with other radars distributed in the large terrain environment, each with similar limited capabilities, we show and illustrate the proposed distributed agents (radars) capability of not only monitoring their respective regions, but also tracking, communicating with neighboring agents, and decision-making, to collaborating span the monitoring the totality of a given large terrain. We show how the neighboring agent radars do not necessarily have to be running on the same computing node. Similarly, as part of the scenario we also consider the case of existence of friendly and unknown forces/stressors targets that are capable of moving throughout the same distributed environment. We show how the proposed algorithms are scalable. They are implemented on the CMINDS High Performance Distributed Computing Engine (HDPC) test-bed taking full advantage of a distributed environment and multiple processing systems.
Date of Conference: 8-10 Nov. 2010