Task parameters analysis in schedule-based timing side-channel attack

Abstract

Recent work has shown that the timing behavior of a real-time system can be utilized by attackers for various adverse purposes via schedule-based timing side-channel attacks. An important assumption in this type of attacks is the prior knowledge of attackers about the task parameter information, including the number of tasks in the system and the period and execution time of each task. The attackers can use such information, together with the execution sequence of the task system, to reconstruct the exact schedule of the tasks and perform various subsequent attacks. In this paper, we show that the schedule-based timing side channel attacks can actually be performed even without knowing the task parameter information in prior. We develop methods to infer the number of tasks in the system and the period and execution time of each task directly from the execution sequence. This removes the task parameter prior knowledge requirement of the attackers and shows greater threats of the schedule-based timing side channel attacks. Both simulation experiments with synthetic task sets and a Zedboard-based evaluation with control system for a 3DOF helicopter are conducted to evaluate the proposed task parameter analysis method.