Affiliations: Jodrey School of Computer Science, Acadia University
Wolfville, Nova Scotia, B4P 2R6 Canada | Department of Computer Science, University of Colorado
at Boulder, CO, USA | Computer Engineering Department, King Fahd University
of Petroleum and Minerals, Dhahran, Saudi Arabia
Note: [] Corresponding author: Jodrey School of Computer Science, Acadia
University, Nova Scotia, Canada B4P 2R6, E-mail: elhadi.shakshuki@acadiau.ca
Abstract: One of the key prerequisite for a scalable, effective and efficient
sensor network is the utilization of low-cost, low-overhead and high-resilient
fault-inference techniques. To this end, we propose an intelligent agent system
with a problem solving capability to address the issue of fault inference in
sensor network environments. The intelligent agent system is designed and
implemented at base-station side. The core of the agent system – problem
solver – implements a fault-detection inference engine which harnesses
Expectation Maximization (EM) algorithm to estimate fault probabilities of
sensor nodes. To validate the correctness and effectiveness of the intelligent
agent system, a set of experiments in a wireless sensor testbed are conducted.
The experimental results show that our intelligent agent system is able to
precisely estimate the fault probability of sensor nodes.
