Multi-Agent System Methodology for Wireless Sensor Networks

Abstract

Recent technological advances in wireless networking, IC fabrication and sensor technology have lead to the emergence of millimeter scale devices that collectively form a Wireless Sensor Network (WSN). The cost of production for a single node has been reduced to less than 1 dollar, paving the way for large scale deployments (millions of nodes per network) of such devices. It would seem that agent technology should be useful for these highly distributed networks in terms of intelligent network management and data harvesting for example. Indeed this has been shown to be the case, however Multi-Agent Systems for WSNs are scarce. One reason for this is the difficulty in the deployment, testing and debugging of a distributed application for these devices due to the minimal (if any) user interfaces they possess. In this paper we present a methodology for the systematic development of a MAS for WSNs that allows for thorough testing and debugging, a luxury not available on current WSN devices. A case study demonstrating the effectiveness of this approach and various methodological tool support are also detailed.