In this paper, we consider how to hide information into finite state machine (FSM), one of the popular computation models. The key advantage of hiding information in FSM is that the hidden information becomes inexpensive to retrieve, yet still hard to remove or delete. This is due to the fact that verifying certain FSM properties is easy, but changing them requires efforts equivalent to redoing all the design and implementation stages after FSM synthesis. We first observe that not all the FSM specifications (or transitions) are needed during the state minimization phase. We then develop a Boolean Satisfiability (SAT) based algorithm to discover, for a given minimized FSM, a maximal set of redundant specifications. Manipulating such redundancy enables us to hide information into the FSM without changing the given minimized FSM. Moreover, when the original FSM does not possess sufficient redundancy to accommodate the information to be embedded, we propose a state duplication technique to introduce additional redundancy. We analyze these methods in terms of correctness, capacity of hiding data, overhead, and robustness against possible attacks. We take sequential circuit design benchmarks, which adopt the FSM model, as the simulation testbed to demonstrate the strength of the proposed information hiding techniques.
展开▼
