Improving Security for Time-Triggered Real-Time Systems with Task Replication

Time-triggered real-time systems achieve deterministic behaviour, making them suitable for safety-critical environments. However, this determinism also allows attackers to finetune attacks after studying the system behaviour through side channels, targeting safety-critical victim tasks. Assuming fault independence, replication tolerates both random and malicious faults of up to f replicas. Yet, directed attacks violate the fault independence assumption. This violation possibly gives attackers the edge to compromise more than f replicas simultaneously, in particular if they can mount the attack from already compromised components. In this paper, we sketch mitigation strategies for time-triggered systems with task replication to withstand directed timing attacks and show preliminary results on their effectiveness and practicality.