Frequency Scaling Based on User Behavior for Smartphone Interactive Application

• Main Authors: Liao, Yu Chiang, 廖毓強
• Other Authors: King, Chung Ta
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/2vxxhs

碩士 === 國立清華大學 === 資訊工程學系 === 105 === With the fast growing technology, the abilities of smartphones become more. In recent years, smartphone is one of the necessities in our daily life. Besides, people tend to use interactive application on their smartphones, such as, social network application, game. People become more care about their smartphone performance and battery life. DVFS (Dynamic voltage and frequency scaling) is an efficient energy saving policy while using smartphones. The policy of frequency scaling usually based on the hardware workload to achieve energy saving. However, this method does not consider the user interaction with the smartphone. In this paper, we collect the user behavior and CPU usage during using Facebook and find out the correlation. Our studies indicate that user behavior would influence the CPU usage during using smartphones. We also implement a user-based policy to scale CPU frequency. Our experiment result shows that it would have 9.2% energy saving and only increase 3.43% CPU usage.