An Energy Characterization Framework for Software-Based Embedded Systems
In this paper, the authors propose an energy characterization framework which helps designers in developing a fast and accurate energy model for a target processor-based system. They use a linear model for energy estimation and they find the coefficients of the model using Linear Programming (LP). They use their approach for estimating the energy consumption of two commercial microprocessors with their on-chip caches and an off-chip SDRAM. Experimental results demonstrate that the error of their technique is on an average 3% and worst case 16% compared to the gate-level estimation results.