Estimation of Sequential Circuit Activity Considering Spatio-Temporal Correlations

In this paper, we present accurate estimation of signal activity at the internal nodes of sequential logic circuits. The methodology is based on stochastic model of logic signals and takes spatial and temporal correlations of logic signals into con~iderai~ionG. iven the State Tranzjition Graph (STG) of a Finite State Machine (FSM), we create an Extended State Transition Graph (ESTG), where the temporal correlations of the input signals are explicitly represented. From the graph we derive the equations to calculiste exact signal probabilities and activities. However, for large circuits the method can be computationally expensive. Therefore, we propose an approximate solution method and a Monte Carlo based approach. The approximate method unrolls the next state logic and calculates the activities using a probabilistic technique which considers spatio-temporal correlations of signals. All Monte Carlo based techniques that have been proposed for combinational circuits so far can not be directly applied to sequential circuits. This is because if the initial transient problem is not well dealt with, the estimated activities for some circuit nodes could be off by more thim 100% in comparison to the exact method. The proposed approach deals with this problem by gaining insight from Markov chain theory. Experimental results show that if temporal and spatial correlations are not considered, the power dissipation determined by the approxi mate method can be off by more than 40%. On the other hand, the results of the approximate method proposed in this paper are within 5% of that of long run simulation. However, for large sequential circuits and for some applications that require the accuracy of the activitit:es of individual nodes, experiments show that the Monte Carlo approach is fast and accurate