ESTIMATORS FOR LOGIC MINIMIZATION AND IMPLEMENTATION SELECTION OF FINITE STATE MACHINES

This paper considers two estimation problems which occur during the implementation design for a finite state machine (FSM). The first is a precise estimation of the reduction of a programmed logic array implementation (PLA) for a FSM by logic minimization. The second concerns selection of implementation alternatives based on such estimations. Estimations give the designer a quick overview of the impact of an optimization method for FSM implementation without running the actual time-consuming algorithms. The method uses curve-fitting on results found in literature for logic minimization preceded by state-assignment. Our estimations correlate by 0.97 to those results.State-graph statistics can also be used for selection of the most profitable optimization from a set of alternatives. We tested selection between a counter based implementation, partial state coding, state-assignment and topological partitioning. The goal is selection of the alternative which has the highest probability to deliver the largest minimization of the FSM. This selection method is also empirically verified by comparing its results with results obtained by running specific optimization algorithms on machines of the MCNC benchmark set.