A Mobile Agent-Based Architecture for Fault Tolerance in Wireless Sensor Networks

Abstract

Wireless Sensor Networks (WSNs) are prone to failures as they are usually deployed in remote and unattended environments. To mitigate the effect of these failures, fault tolerance becomes imperative. Nonetheless, it remains to be a second tier activity—it should not undermine the execution of the mission oriented tasks of WSNs through overly taxing their resources. We define architecture for fault tolerance in WSNs that is based on a federation of mobile agents that is used both for diagnostic intelligence and repair regimen, focusing on being lightweight in energy, communication and resources. Mobile agents are classified here as local, metropolitan, and global, providing fault tolerance at node, network and functional levels. Interactions between mobile agents are inspired by honey bee dance language that builds on semantics of errors classification and their demographic distribution. Our quantitative modeling substantiates that the proposed fault tolerance framework mandates minimalist communication through contextualized bee-inspired interactions, achieving adaptive sensitivity, and hysteresis-based stability