Efficient statistical approach to estimate power consideringuncertain properties of primary inputs

摘要

Power dissipation in complementary metal-oxide-semiconductorn(CMOS) circuits is heavily dependent on the signal properties of thenprimary inputs. Due to uncertainties in specification of suchnproperties, the average power should be specified between a maximum andna minimum possible value. Due to the complex nature of the problem, itnis practically impossible to use traditional power estimation techniquesnto determine such bounds. In this paper, we present a novel approach tonaccurately estimate the maximum and minimum bounds for average powernusing a technique which calculates the sensitivities of average powerndissipation to uncertainties in specification of primary inputs. Thensensitivities are calculated using a novel statistical technique and cannbe obtained as a by-product of average power estimation using MontenCarlo-based approaches. The signal properties are specified in terms ofnsignal probability (probability of a signal being logic ONE) and signalnactivity (probability of signal switching). Results show that thenmaximum and minimum average power dissipation can vary widely if thenprimary input probabilities and activities are not specified accurately