Implementation of Sequential Detection of Replay Attacks for a Cloud Native Controller

This paper focuses on the investigation of a sequential detection policy for combating the effect of the replay attack on a cloud-native controller. In reply attacks, an external attacker records true signals for some time duration and replays them later. The effect of such an attack is mitigated by adding random signals to the optimal control signal before the actuation process, which is known as the watermarking policy. We study the effectiveness of this joint scheme of sequential detection with watermarking in a cloud-native controller with varying levels of delay introduced in the communication network between the cloud server and the physical plant. Our study shows that the scheme performs well for the cloud-native controller and that the scheme is able to mitigate the effect of the replay attack when the control signals arrive in a timely fashion, i.e., without significant delay from the controller to the physical plant.