Optimal Machine Learning Based Intrusion Detection System in Wireless Sensor Networks for Surveillance Applications

Abstract

Security is considered as a major design issue in wireless sensor network (WSN) and can be solved by the use of intrusion detection systems (IDS). In this view, this paper devises a new k-means clustering with optimal support vector (KM-OSVM) based IDS for WSN. The KM-OSVM model incorporates preprocessing, clustering, classification, and parameter tuning. Primarily, data preprocessing and K-means clustering technique are applied to group the data instances into a set of clusters. Besides, SVM based classification technique is employed to allot class labels, and the parameters in SVM are optimally adjusted by the use of crow search optimization (CSO) algorithm, shows the novelty of the work. The experimental outcome of the KM-OSVM model is examined using UNSW-NB15 and CICIDS2017 datasets. The obtained outcomes demonstrated that the KM-OSVM model ensured better performance with the maximum accuracy of 95.12% and 98.98% respectively. Therefore, the KM-OSVM model can be employed as an effective tool to achieve security in the resource constrained WSN.