Authors: Song Shen, Gregory M.P. O'Hare
Addresses: School of Computer Science and Informatics, University College Dublin, Belfield, Dublin 4, Ireland. ' School of Computer Science and Informatics, University College Dublin, Belfield, Dublin 4, Ireland
Abstract: Large-scale wireless sensor applications require that the Wireless Sensor Network (WSN) be sufficiently intelligent to make decisions or tradeoffs within a partially uncertain environment. Such intelligence is difficult to model because the sensor nodes are resource-bounded, especially on energy supply. Within this paper, we model a WSN as a two-level multiagent architecture and introduce the concept of energy-aware and utility-based Belief, Desire, Intention (BDI) agents. We subsequently formalise the fuzzy-set-based belief-generating and theoretically deduce the rules of the meta-level reasoning of a WSN. We further apply a practical reasoning algorithm which enables cost awareness and status awareness within a WSN. A Tileworld-based simulation and subsequent comparison of the effectiveness of bold, cautious and energy-aware and utility-based agents demonstrates that an energy-aware and utility-based BDI agent outperforms other traditional agents, especially when the environment is highly dynamic.
Keywords: BDI agents; energy aware agents; utility based agents; meta-level reasoning; wireless sensor networks; multi-agent systems; agent-based systems; wireless networks; WSNs; fuzzy sets; fuzzy logic; belief generation; simulation.
International Journal of Sensor Networks, 2007 Vol.2 No.3/4, pp.235 - 245
Available online: 11 Apr 2007 *Full-text access for editors Access for subscribers Purchase this article Comment on this article