An Optimized SVM based Fault Diagnosis Scheme for Wireless Sensor Networks

Recently, Wireless Sensor Networks are engaged to solve several real life problems such as environmental monitoring, home automation, medical monitoring. In such applications, small sensor nodes are susceptible to failure, due to the hardware failures, software failure and energy exhaustion. Faulty sensor nodes produce erroneous measurements that can reduce performance of the WSNs. Therefore, fault diagnosis and detection is one of the most critical issues in WSNs. This paper proposes an optimized Support-Vector Machine (SVM) based fault diagnosis scheme for wireless sensor networks. In the proposed scheme, Grey Wolf Optimization (GWO) based SVM classifier is used to detect fault status of the deployed sensor nodes. In addition, a cluster based topology is proposed for energy saving purpose. We perform massive simulation on the proposed fault detection method in different network settings. The simulation outcomes are compared with the existing schemes to validate the potency of the proposed fault detection method under different performance criteria.