Statistical Modeling of Glitching Effects in Estimation of Dynamic Power Consumption
Authors:
- Michael MeixnerMMMichael Meixner
- Tobias G. NollTNTobias G. Noll
- 1 January 2014
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 415-420
Abstract
A new approach for high-level power estimation is presented which considers the increasingly important effect of glitching. The approach is based on modeling of functional building blocks by pre-characterization. This way the accuracy of modeling at the circuit level can be preserved to higher levels of abstraction. New metrics employing probability functions are applied to quantitatively describe the glitch generation and transfer characteristic as well as the power induced by glitching. These empirical probability functions are approximated using superimposed standard distributions in order to significantly reduce the number of involved parameters. In this contribution it is investigated what maximum accuracy can be achieved by this approach when tolerating a complex modeling step. As the models resulting from pre-characterization can be reused in other designs, larger modeling efforts are acceptable. Based on the evaluation of the essential part of array multipliers it is shown that the presented approach is suitable to predict the glitching part of power consumption with good accuracy.Keywords
This publication has 12 references indexed in Scilit:
- Process Variations and Probabilistic Integrated Circuit Design2012
- A theoretical probabilistic simulation framework for dynamic power estimation2011
- Statistical full-chip dynamic power estimation considering spatial correlations2011
- Dynamic power estimation for deep submicron circuits with process variation2010
- A Timing-Dependent Power Estimation Framework Considering CouplingIEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2009
- A New Statistical Approach for Glitch Estimation in Combinational Circuits2007
- High-level macro-modeling and estimation techniques for switching activity and power consumptionIEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2003
- Power modeling for high-level power estimationIEEE Transactions on Very Large Scale Integration (VLSI) Systems, 2000
- Architectural power analysis: The dual bit type methodIEEE Transactions on Very Large Scale Integration (VLSI) Systems, 1995
- Transition density: a new measure of activity in digital circuitsIEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 1993