Toward Energy Profiling of Connected Embedded Systems

Abstract

International audienceA huge number of connected objects are expected to be deployed over the coming years in various areas of everyday life. Many of these objects are energy-constrained and depend on a battery. Thus, energy is a critical resource that limits a large scale deployment, and greatly complicates the development of the embedded software on these objects. Hence, the ability to measure and finely profile the power consumption of such devices, and correlate it with the on-board application is a big challenge to improve the software development. Furthermore, common energy patterns can be extracted from the collected energy figures in order to provide guidelines allowing a proactive energy-based development. In this paper, we present an ongoing work about a lightweight framework for energy profiling of embedded applications source code at a functional granularity. It is driven by an on-line hardware-based measurement technique permitting to gather accurate energy figures. The framework is integrated into an energy-centric iterative development cycle allowing fast revalua-tions of the energy consumed by the targeted functions after each source code modification. Afterwards, we describe our future works about overcoming an unlocked state of art issue relative to asynchronous energy consumption profiling

Automatic Inference of Energy Models for Peripheral Components in Embedded Systems

“…We believe that this schema allows the developer to get easily aware of the energy consumption, induced by each action performed by the embedded peripheral component. This work is based on an existing energy profiling framework already described and presented in a precedent research article [3]. This paper is organized as follows.…”

“…The energy-based iterative development cycle, aiming to enhance and to simplify the software development for energyconstrained embedded applications[3].…”