Mobile and Ubiquitous Systems: Computing, Networking, and Services. 10th International Conference, MOBIQUITOUS 2013, Tokyo, Japan, December 2-4, 2013, Revised Selected Papers

Research Article

Evaluation of Energy Profiles for Mobile Video Prefetching in Generalized Stochastic Access Channels

Download
474 downloads
  • @INPROCEEDINGS{10.1007/978-3-319-11569-6_17,
        author={Alisa Devlic and Pietro Lungaro and Zary Segall and Konrad Tollmar},
        title={Evaluation of Energy Profiles for Mobile Video Prefetching in Generalized Stochastic Access Channels},
        proceedings={Mobile and Ubiquitous Systems: Computing, Networking, and Services. 10th International Conference, MOBIQUITOUS 2013, Tokyo, Japan, December 2-4, 2013,  Revised Selected Papers},
        proceedings_a={MOBIQUITOUS},
        year={2014},
        month={12},
        keywords={Energy profiles Stochastic access channel Mobile video prefetching},
        doi={10.1007/978-3-319-11569-6_17}
    }
    
  • Alisa Devlic
    Pietro Lungaro
    Zary Segall
    Konrad Tollmar
    Year: 2014
    Evaluation of Energy Profiles for Mobile Video Prefetching in Generalized Stochastic Access Channels
    MOBIQUITOUS
    Springer
    DOI: 10.1007/978-3-319-11569-6_17
Alisa Devlic1,*, Pietro Lungaro1,*, Zary Segall1,*, Konrad Tollmar1,*
  • 1: Royal Institute of Technology (KTH)
*Contact email: devlic@kth.se, pietro@kth.se, segall@kth.se, konrad@kth.se

Abstract

This paper evaluates the energy cost reduction of Over-The-Top mobile video content prefetching in various network conditions. Energy cost reduction is achieved by reducing the time needed to download content over the radio interface by prefetching data on higher data rates, compared to the standard on demand download. To simulate various network conditions and user behavior, a stochastic access channel model was built and validated using the actual user traces. By changing the model parameters, the energy cost reduction of prefetching in different channel settings was determined, identifying regions in which prefetching is likely to deliver the largest energy gains. The results demonstrate that the largest gains (up to 70 %) can be obtained for data rates with strong correlation and low noise variation. Additionally, based on statistical properties of data rates, such as peak-to-mean and average data rate, prefetching strategy can be devised enabling the highest energy cost reduction that can be obtained using the proposed prefetching scheme.