Finite state machine decomposition by transition pairing

The authors develop a method based on the premise that optimal state assignment corresponds to finding an optimal general decomposition of a finite state mechanism (FSM). They discuss the use of this approach for encoding state transition graphs extracted from logic-level descriptions. The notion of transition pairing is used to decompose a given FSM into several submachines such that the state assignment problem for the submachines is simpler than the original problem, attempting to avoid compromising the optimality of the solution. A novel decomposition algorithm that can decompose a FSM into an arbitrary number of submachines and a novel constraint satisfaction algorithm to encode the different submachines are given. Experimental results validate the use of decomposition-based techniques to solve the encoding problem.<>