Detection of Intrusion using Enhanced Machine Learning Model in SCADA Wireless Network

Abstract

The remote communication and its control purposes are highly integrated and the control through the wireless network is monitored by Supervisory Control and Data Acquisition (SCADA) systems. The attacks are isolated based on the optimal feature selection that is extracted from the sensor data. The cluster between the matrix and the optimal features are extracted and labeled. During clustering, the initial processing of attacks is removed and it is performed by the Mean shift clustering algorithm. The irrelevant features from the clustered data are concealed using the Intrusion detection system based Enhanced Cuckoo Search optimization algorithm (IDS-ECSO) and it is used to select the best features. The relevancy vector is used to classify the attacks and it is performed using Genetic Machine Learning based Neural Network (GML-NN). The classification results of the SCADA data set with the proposed method are analyzed and compared with the other methods.