An Energy Efficient Event Detection Classifier in Wireless Sensor Network Using Support Vector Machine

Wireless sensor network (WSN) comprises of nodes that are spatially distributed to monitor the environments and detect the events accordingly. Correlated Data Gathering (CDG) in wireless sensor network used Adaptive and Distributed Routing (ADR) algorithm for correlated data gathering in order to minimize the total energy consumption. Though energy consumption was reduced in the network, energy delay tradeoff occurred while securing data in sensor network was high. Energy-efficient and High-accuracy (EEHA) scheme provided a secured data aggregation technique using an aggregation tree which provided privacy but the energy consumption was high during event detection. Recoverable Concealed Data Aggregation (RCDA) for Data Integrity recovered all sensing data events even when the data were aggregated, by reducing the transmission overhead but with higher energy ratio. To develop an energy efficient Event Detection Classifier in wireless sensor network, a predetermined event weight based on Support Vector Machine (EDC-SVM) is proposed in this paper. The EDC-SVM initially identifies the weight of the events for effective classification using SVM with minimal energy consumption. EDC-SVM uses Doppler Effecting method for recovering all sensing data events with minimal energy. The task of Doppler Effecting method in EDC-SVM is to detect the periodic events of moving objects (i.e.,) sensor nodes to reduce the classification time of sensor events. With the minimal time on classification, the energy delay tradeoff is overcome in EDC-SVM. Furthermore, with the application of an event section key generation in EDC-SVM, reduces the energy consumption during the generating of section key and broadcast the notification to the sensor nodes within the section. EDC-SVM with event section key generation improves the security level on object collection. Experimental work is carried out on the factors such as classifier rate, security level, and energy consumption rate.