Usage pattern-based prefetching: quick application launch on mobile devices

  • Authors:

  • Hokwon Song;Changwoo Min;Jeehong Kim;Young Ik Eom

  • Affiliations:

  • Samsung Electronics Co., Ltd., Suwon, Korea,School of Information and Communication Engineering, Sungkyunkwan University, Suwon, Korea;Samsung Electronics Co., Ltd., Suwon, Korea,School of Information and Communication Engineering, Sungkyunkwan University, Suwon, Korea;School of Information and Communication Engineering, Sungkyunkwan University, Suwon, Korea;School of Information and Communication Engineering, Sungkyunkwan University, Suwon, Korea

  • Venue:
  • ICCSA'12 Proceedings of the 12th international conference on Computational Science and Its Applications - Volume Part III
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

The startup time of applications is very important as a user perspective performance. If page faults occur frequently in the startup time, the user experience is subjected to an adverse effect. To reduce page faults, the prefetching scheme is used in the traditional OS. Previous studies proposed various schemes, but the most research was conducted for desktop PCs or special embedded devices. We propose the usage pattern-based prefetching scheme which is suitable to mobile devices. Therefore, this paper focuses on the user's applications usage patterns and the improvement of the startup time of application on mobile devices. To inspect the usage patterns, we collect the dataset of the application usage and then analyze collected data. Additionally, considering mobile devices which have relatively poor hardware resources, the lightweight prediction model is employed in the new scheme. The proposed scheme is implemented on both Android 2.2 and Linux kernel 2.6.29. It is tested on the emulator and evaluated by using the dataset. The startup time is improved about 5%, and the accuracy of the prediction is shown up to 59% for the practical dataset.