Prediction of abnormal temporal behavior in real-time systems.

Ensuring security in real-time and safety-critical systems is becoming extremely challenging, in particular due to the increasingly connectivity of these systems, such as in emerging autonomous vehicles that are subject to new and higher number of security attacks. The main characteristics of real-time systems is that they have strict timing constraints. These constraints must be met in order to ensure the correctness of the system. In this paper, we use the temporal properties derived from the timing analysis of the system required for safety, in order to detect the misbehavior of the system and improve its corresponding security features. For this, we define temporal boundaries which are used to detect the temporal off-nominal behavior of running applications. Moreover, we present a prediction scheme which uses the designated configuration in real-time systems.