TaskShuffler: A Schedule Randomization Protocol for Obfuscation against Timing Inference Attacks in Real-Time Systems

Abstract

The high degree of predictability in real-time systems makes it possible for adversaries to launch timing inference attacks such as those based on side-channels and covert-channels. We present TaskShuffler, a schedule obfuscation method aimed at randomizing the schedule for such systems while still providing the real-time guarantees that are necessary for their safe operation. This paper also analyzes the effect of these mechanisms by presenting schedule entropy - a metric to measure the uncertainty (as perceived by attackers) introduced by TaskShuffler. These mechanisms will increase the difficulty for would-be attackers thus improving the overall security guarantees for real-time systems.