Estimating power consumption of mobile embedded software based on behavioral model

The importance of software that consumes low power in embedded application is revealed again. The existing researches to analyze power consumption of embedded software have been focused on instructions or source-codes level. However, these approaches have shortcomings that the analysis time is long and the power reduction effect is not great. This paper proposes a power consumption analysis technique in more abstract model level than source-code. When our proposed technique compared with the existing source-code analysis, the estimating result of power consumption shows the deviation within average 10% and the analysis time is reduced more than 80%. Our proposed technique can be also applied very usefully to select design decisions that consume low power in the software design phase.