ANFIS based cyber physical attack detection system

Cyber-Physical Systems integrate computer networks and physical process. The security aspect of such system is one of the challenges for the researchers. In this paper, we present an adaptive neuro-fuzzy inference system based Intrusion detection system for the cyber-physical system. The proposed intelligent intrusion detection system is a security tool that allows authorized access to the desired system by examining the network traffic and the previous record. The standard KDD Cup 99 dataset is used to identify different types of intrusion signatures and features. Chi-Square test and Correlation matrix is used to select relevant features and then the sets of features help to classify the attack types. Performance evaluation shows that our proposed work improves the accuracy for classification of different attack types.