Fine-Grained Energy Estimation and Optimization of Embedded Operating Systems

Embedded operating systems (EOS) manage the resources of the system and control device operations, and play an important role on optimizing system energy consumption. This paper proposes a new approach to estimate and optimize the energy consumption of the EOS and the applications at a fine-grained level. The approach is based on a micro-architectural power model and a new estimation model for operating system energy consumption. We apply the approach to an Intel Strong-Arm architecture platform running embedded Linux 2.4.18, analyzing its energy characteristics and also trying to optimize energy of the applications on it based on the analyzing results. The experiments demonstrate that the approach can identify energy consumption of fine-grained software components correctly and be used to optimize the energy consumption of EOS and applications.