Efficient Finite State Machine Encoding for Defending Against Laser Fault Injection Attacks

Finite State Machines (FSMs) are widely used to implement complex computation sequences and communication protocols. An FSM may consist of different states with different privilege levels, such as protected and non-protected. Ideally, switching from a non-protected state to a protected state should involve an authorization transition. However, with Laser-based Fault injection (LFI), an attacker can bypass authorization by flipping bits in the FSM’s state vector. In order to mitigate LFI vulnerability, one can encode the FSM states with the objective of maintaining a large Hamming Distance (HD) between each pair of states. The existing FSM encoding algorithms are either very slow, rely on the user’s mathematical ability to manually generate certain state encodings, or lead to unacceptable area overhead. In this paper, we propose an automated framework for generating FSM encodings to defend against LFI attacks. The proposed technique is a fast linear code-based heuristic to produce area-efficient results. We also propose an application-specific simplification to further reduce the area overhead. Experimental results demonstrate that our proposed method is several orders-of-magnitude faster than the state-of-the-art approaches. Our approach also significantly reduces the state code length (50% on average) compared to state-of-the-art approaches.