Because mobile devices, like smartphones, are usually powered with small-size batteries that have limited capacity and life, hardware engineers need to be smart at designing energy-efficient systems to run the applications with minimal energy. On the other hand, software engineers need to be aware of the energy consumption behaviors of the hardware components. Therefore, analyzing and estimating the energy consumption of the devices during runtime is crucial. There are usually two ways of doing this. The first strategy is hardware based, which is to measure the current and power consumption with instruments, like a source meter. The limitation of this solution is that it is not portable, and it requires opening the devices physically during the measurement. A better solution would be software based, which is called power/energy profiling. The idea is to characterize power/energy at the software level based on the power models that are trained using power measurements and system logs.

In this paper, Hoque and colleagues provide a comprehensive survey of the existing software-based energy profiling solutions for analyzing and eliminating smartphone energy consumption. The survey covers a broad range of solutions, from the basic ones that are just able to report total system power to the most advanced ones that are able to provide the energy consumption profile on the program code. It also covers several energy diagnosis engines, which can detect abnormal energy use by different applications and analyze the reasons for this energy use so that developers are aware of the behaviors and can make decisions on how to optimize the applications. The survey starts by discussing the necessary steps to construct an energy profiler and then presenting different methods for each step and substep. This makes the paper very clear and easy to follow.

This paper provides a very complete design space analysis by comparing each solution and analyzing the tradeoffs. It is a very helpful guide for researchers who have just started working with software-based energy profilers and want to learn the field. Also, it will help software developers choose the right energy profilers based on different applications or power/energy requirements.