Deep learning algorithm based cyber-attack detection in cyber-physical systems-a survey

Over the last years, cyber-attack detection and control system design has become a significant area in cyber-physical systems (CPSs) due to the rapid growth of cyber-security challenges via sophisticated attacks like data injection attacks, replay attacks, etc. The effect of different attacks may provide system failure, malfunctioning, etc. As a result, an improved security system may require to implement the cyber defense system for upcoming CPSs. The different deep learning algorithm based cyber-attack detection schemes have been designed to detect and mitigate the different types of cyber-attacks through CPSs, smart grids, power systems, etc. This article presents a detailed survey of various deep learning algorithms proposed for CPSs to achieve cyber defense. At first, different algorithms developed by previous researchers are studied in detail. Then, a comparative analysis is carried out to know the limitations in each algorithm and provide a suggestion for further improvement of CPSs with more efficiently.