Exploiting structural knowledge using network description language and causal models for fault diagnosis in wireless sensor networks

Abstract

Fault Diagnosis is an essential management task for any telecommunication network and it is even more crucial for Wireless Sensor Networks due to their dynamic nature. Based on Agent Technology, this paper presents an architecture that combines different network and diagnosis models to carry out a Fault Diagnosis process: a Causal Model to relate fault root causes with their symptoms and a Structural Model to define the network and its properties. The behaviour of the proposed agent is driven by an Expertise Model which handles the different phases, tasks and problems which can be found in any diagnosis process. The proposed approach has been evaluated in a simulated environment with emulated MICAz devices for a motion detection application.