A Simulation Based Approach to EstimateEnergy Consumption for Embedded Processors

Abstract: Embedded systems have entered a new era in which system designers have to consider more and more strict energy consumption constraints. This thesis reviewsprevious studies of the processor energy consumption estimation. Particularly, wefocus on instruction-level energy consumption for embedded processors and explorethe energy consumption model for manycore architecture in real traffic pattern.The purpose of this thesis project is to estimate energy consumption and constructan energy model using an instruction-set simulator for embedded processors. OpenVirtual Platforms (OVP) and Epiphany Single Core Simulator (ESCS) are used toobtain an instruction sequence for a given software. Then, the functionality of energyconsumption estimation is integrated into OVP.Our energy consumption estimation approach categorizes instructions in fourgroups and uses base energy cost of each category to calculate the total energyconsumption for an application that runs on Epiphany core and ARM Cortex M0.The energy consumption estimation for Epiphany core has been successfully testedby BEEBS benchmarks. Details of the derivation process, results and analysis ofenergy consumption estimation methods are provided.