No abstract

Context SCADA systemsSCADA which stands for Supervisory Control And Data Acquisition, are Industrial Control Systems (ICS) used for centralized data acquisition and control of geographically dispersed assets. Those systems are used in water distribution, wastewater treatment, power transmission and distribution, oil and gas pipelines, public transportation systems, etc. [145]. The integration of SCADA systems to the management of industrial systems not only helps improve performance, but also reduces operating costs [21].Historically, no consideration was given to security at the time of the design of SCADA systems, but engineers were more focused on practical aspects such as availability, reliability, and performance of physical processes [91]. Designers relied upon two forms of protection i.e. air gap [22] and security through obscurity. The former was based on the fact that SCADA networks are physically isolated from other networks making any attack difficult, while the latter relied on the presumption that information about SCADA systems are not available to public, thereby making them secured [94]. Attacks on SCADA systemsNowadays, modern SCADA systems use Commercial-Off-The-Shelf (COTS) hardware, software, standard communication technologies such as TCP/IP and ETH-1 In the next chapter, we provide the reader some insights of deep learning fundamentals, common architectures, and their unsupervised feature learning capability that could be used for anomaly detection purpose.