Mobile Computing, Applications, and Services. 5th International Conference, MobiCASE 2013, Paris, France, November 7-8, 2013, Revised Selected Papers

Research Article

Reconciling Cloud and Mobile Computing Using Activity-Based Predictive Caching

Download56 downloads
  • @INPROCEEDINGS{10.1007/978-3-319-05452-0_11,
        author={Dawud Gordon and Sven Frauen and Michael Beigl},
        title={Reconciling Cloud and Mobile Computing Using Activity-Based Predictive Caching},
        proceedings={Mobile Computing, Applications, and Services. 5th International Conference, MobiCASE 2013, Paris, France, November 7-8, 2013, Revised Selected Papers},
        proceedings_a={MOBICASE},
        year={2014},
        month={6},
        keywords={predictive caching activity recognition mobile cloud computing connectivity prediction mobile apps},
        doi={10.1007/978-3-319-05452-0_11}
    }
    
  • Dawud Gordon
    Sven Frauen
    Michael Beigl
    Year: 2014
    Reconciling Cloud and Mobile Computing Using Activity-Based Predictive Caching
    MOBICASE
    Springer
    DOI: 10.1007/978-3-319-05452-0_11
Dawud Gordon1,*, Sven Frauen1,*, Michael Beigl1,*
  • 1: Karlsruhe Institute of Technology (KIT)
*Contact email: Dawud.Gordon@kit.edu, Sven.Frauen@kit.edu, Michael.Beigl@kit.edu

Abstract

Cloud computing has greatly increased the utility of mobile devices by allowing processing and data to be offloaded, leaving an interface with higher utility and lower resource consumption on the device. However, mobility leads to loss of connectivity, making these remote resources inaccessible, breaking that utility completely during offline periods. We present a concept for reconciling the fragile connectivity of mobile devices with the distributed nature of cloud computing. We predict periods without connectivity on the mobile devices before they occur and cache process states for applications running on distributed cloud back-ends. The goal is to maintain partial or full utility during offline periods, and thereby to enable an improved user experience. We demonstrate prediction must include real-time behavioral information in addition to location and temporal models. The approach is implemented for mobile phones which learn to quantify human behavior using activity recognition, and then learn patterns in that behavior which lead to disconnectivity. We evaluate it for a streaming music scenario, where data is cached before the user goes offline, allowing seamless playback. The results show that theoretically we can successfully predict 100% of disconnection events on average 8 minutes in advance (std. dev. 46 secs.) with minimal false-positive caching in this scenario, although in the wild these events could prove more difficult to predict.