Role swapping in multi-agent sensor webs
Rónán Mac Ruairí, Mark T. Keane · European Conference on Artificial Intelligence · 2004
This paper reports on the application of multi-agent systems to the design of sensor networks. In advance of the current hardware/software capabilities of these sensor networks, we simulate a sensing network handling the detection of a contaminant, in a hostile environment, where the effective life of the network has to be optimised. By casting each sensor as an agent that can swap its role as conditions in the network change we show that networks using role-swapping, as opposed to fixed roles, can be much more effective over time. The emergent behaviour, that arises from economical local interactions, also becomes more effective as the environment for the network becomes progressively harsher.