검색 상세

Design of Thermal Mitigation Algorithm based on Watching Videos in Smartphone

스마트폰에서 동영상 시청을 기반으로 한 발열 경감 알고리즘 설계

초록/요약

Initially, mobile devices were used only for text messages or calls. However, with the recent development of mobile communication and processors, users can enjoy various contents that require high-performance such as watching videos and playing games. However, to use such content required for high performance, the device must operate with a high operating frequency, which causes problems such as increased power consumption and thermal. Also, unlike a computer environment, it is difficult to use a separate thermal mitigation technique for mobile devices, and thermal mitigation is managed through software. Therefore, focusing on the recent increase in using video content, a study was conducted to propose a thermal mitigation algorithm of CPU, one of the software thermal mitigation techniques. In this paper, we conducted an experiment by selecting YouTube, Netflix, Nav-er TV, and Afreeca TV, among the video applications that users generally use. We analyzed the CPU operating pattern that appears when watching a video on a smartphone. We then confirmed that there were operating frequencies used, as well as inefficient operating frequencies, depending on the number of operating cores. In addition, since the application plays the video as streaming, we extracted and analyzed the data throughput that appears when watching a video, then confirmed the characteristics and patterns of data throughput. The thermal mitigation algorithm proposed in this paper is first classified into two types based on characteristics and the pattern of data throughput. Then, based on the previously analyzed CPU operating pattern, the algorithm was designed to operate at the optimum operating frequency, by setting the minimum and maximum allowed operating frequency, which was in accordance with the number of operating cores. The target device used in the experiment is LG G3, and the application processor, which is a Snapdragon 801 composed of a quad-core CPU, was equipped. In the experiment, the video was played in each application for about 40 minutes, the power consumption was measured by the battery consumption consumed by the smartphone, the thermal was measured using the temperature sensor built into the application processor, and the performance was measured by FPS (Frame Per Second). As a result of comparing the default thermal mitigation algorithm and the proposal thermal mitigation algorithm, the overall average temperature decreased by about 2.5 ℃, the battery consumption decreased by about 1 %, and the FPS improved by 4.56 on average. As a result, the proposal thermal mitigation algorithm was designed to operate at a frequency optimized for watching the video, decreasing power consumption, thermal, and increasing the overall performance. Through this, it is meant that it suggested that the default thermal mitigation algorithm can be improved if it is designed to operate at a frequency suitable for the target task.

more