Since smartphones are battery-driven, it is important to know how power-efficiently and how long they are used in a limited battery capacity. As smartphones become popular and their usage time increases, many low-power techniques have been studied to use smartphones' batteries efficiently. However, most low-power techniques are only interested in techniques that reduce power purely, and do not care about the feasible time of the current workload based on the battery remaining time. In this paper, we propose a method to predict the accurate power consumption of recent smartphones for development of efficient low power techniques. Our method estimates the battery remaining time to predict the viable time of the running workload with the remaining smartphone battery. In detail, it predicts the overall power consumption for a smartphone by using power models for a CPU, a GPU, and a display that have a high proportion of energy consumption in smartphones. Through experiments, we build a prediction model that calculates battery consumption by using the predicted power consumption according to CPU/GPU phase of the workload used in the experiment. Consequently, we complete a technique to predict battery consumption and remaining time for arbitrary workloads.
ACKNOWLEDGMENT This work has been supported by the Future Combat System Network Technology Research Center program of Defense Acquisition Program Administration and Agency for Defense Development.(UD109 033ED)