Low-power driven state assignment of finite state machines

We present a new state assignment model for minimizing power dissipation in Finite State Machines (FSMs). Our model takes into account power dissipation in terms of transition densities during state assignment process such that the number of transitions of the state encoding bits is minimized. Given the set of input signal probabilities, we compute not only the transition probabilities but also the exact state probabilities, the fractions of time for FSMs being in certain states, to be used in an objective weight function depending on encodings of states in FSMs. For every benchmark example circuits, our model always generates the least number of transitions among several state assignment models that confirms the effectiveness of our model.