Multilayer perceptron with binary weights and activations for intrusion detection of Cyber-Physical systems
Paul Vincent Alpano, Jhoanna Rhodette Pedrasa, Rowel Atienza · 2017
Intrusion detection of Cyber-Physical systems has been a major concern since there are multiple ways of entry for cyber-attacks. Methods have been proposed to apply intrusion detection techniques to CPS, but most of them are host-centric. Problems arise for CPS with host-centric intrusion detection if the actual host is the one compromised. We introduce a method of having intrusion detection system for other low-powered devices of CPS through binary weights and activations of Multilayer Perceptron neural network. Our multilayer perceptron was trained, validated, and tested through the KDD '99 dataset and produced significant results which is discussed in the paper.