Strati ed Random Sampling for Power

| In this paper, we present new statistical sampling techniques for performing power estimation at the circuit level. These techniques rst transform the power estimation problem to a survey sampling problem, then apply stratiied random sampling to improve the eeciency of sampling. The stratiication is based on a low-cost predic-tor, such as zero delay power estimates. We also propose a two-stage stratiied sampling technique to handle very long initial sequences. Experimental results show that the ee-ciency of stratiied random sampling and two-stage stratiied sampling techniques are 3-10X higher than that of simple random sampling and the Markov-based Monte Carlo simulation techniques.