fiee 15(5): e3

Research Article

Tradeoff Analysis for Mobile Cloud Offloading Based on an Additive Energy-Performance Metric

Download448 downloads
  • @ARTICLE{10.4108/icst.valuetools.2014.258222,
        author={Huaming Wu and Katinka Wolter},
        title={Tradeoff Analysis for Mobile Cloud Offloading Based on an Additive Energy-Performance Metric},
        journal={EAI Endorsed Transactions on Future Intelligent Educational Environments},
        volume={1},
        number={5},
        publisher={EAI},
        journal_a={FIEE},
        year={2015},
        month={2},
        keywords={energy-performance tradeoff, queueing model, offloading policies, mobile cloud computing},
        doi={10.4108/icst.valuetools.2014.258222}
    }
    
  • Huaming Wu
    Katinka Wolter
    Year: 2015
    Tradeoff Analysis for Mobile Cloud Offloading Based on an Additive Energy-Performance Metric
    FIEE
    EAI
    DOI: 10.4108/icst.valuetools.2014.258222
Huaming Wu1,*, Katinka Wolter1
  • 1: Freie Universit√§t Berlin
*Contact email: wuhuaming2006@gmail.com

Abstract

Mobile offloading migrates heavy computation from mobile devices to powerful cloud servers. It is a promising technique that can save energy of the mobile device while keeping job completion time low when cloud servers are available and accessible. The benefit obtained by offloading greatly depends on whether it is applied at the right time on the right way. In this paper, we use queueing models to minimize a weighted sum of energy consumption and performance expressed in the Energy-Response time Weighted Sum (ERWS) metric. We consider different offloading policies (static and dynamic), where arriving jobs are processed either locally or remotely in the cloud. Offloading can be performed via WLAN or via a cellular network. The transmission techniques differ in energy requirement and speed. We find that the dynamic offloading policy derived from the tradeoff offloading policy (TOP) outperforms other policies like the random selection of transmission channel by a significant margin. This is because the dynamic offloading policy considers the increase in each queue and the change in metric that newly arriving jobs bring in should they be assigned to that queue. The ERWS metric can be reduced more by considering either energy consumption or response time and it is minimal when optimising only energy consumption.