Book chapter 9: Applying Machine Learning and Deep Learning Algorithms for the Detection of Physical Anomalies in Critical Water Infrastructures

Industrial Control Systems Security implies the safekeeping and protection of such systems, as well as all the software and hardware used by them. Restrict logical and physical access to the ICS devices and networks, securing all individual components of the ICS or avoid unauthorized changes of data are some the main objectives of the ICS security, however, knowing when you are being victim of an attack is more and more important. For this reason, threat detection in industrial infrastructures represents an actual and worthwhile research topic. In this chapter, we present two security tools developed in the STOP-IT project that applyMachine Learning and Deep Learning algorithms to detect abnormal behaviours or situations that could become physical threats for aWater Infrastructure. A device able to detect the presence of a person in a room or a delimited area by analysing the reflection of Wi-Fi signals in human body and a system able to identify intrusions and abnormal movements or behaviours around the water facility by using improved computer vision techniques.