Energy Aware Mobile Cloud Computing

点击翻译按钮获取中文摘要

Mobile Internet communications has seen enormous growth in the past two decades, and is continuing to have a profound impact on many aspects of our personal and business lives. This growth is being fuelled by the proliferation of Internet-connected cellphones, laptops, and other wireless devices. Over this time, "mobile cloud computing" has emerged as a key driver of mobile Internet applications. In mobile cloud computing, wireless devices access online Internet-hosted servers that manage, store, and process data, rather than doing so locally. Since mobile devices tend to be far more resource limited than their infrastructure counterparts, there are many advantages to mobile cloud computing, which is fuelling its explosive growth. Mobile cloud traffic is expected to increase by more than an order of magnitude in the next five years alone.As cloud computing grows, the nature of mobile-to-cloud networking will evolve, and will include more tightly coupled mobile-to-cloud interactions. This includes "computation offloading", where tasks that would otherwise be executed locally, are offloaded to cloud servers for remote execution. There are many benefits to this, including reduced handset energy use and the execution of more sophisticated applications that would be impossible otherwise. Computation offloading promises to enable a new generation of mobile Internet applications and services.There are some key technical issues to overcome in order to exploit the full promise of computation offloading. This is the objective of the proposed research. Recent experiments have shown for example, that computation offloading performs very well under ideal wireless and cloud server conditions, which is rarely the case in existing cellular and Wi-Fi deployments. In addition, the current state-of-the-art has tended to focus on isolated mobile-to-cloud interactions, and has tended not to incorporate the effects of resource contention, practical wireless impairments, and infrastructure placement. The ultimate success of computation offloading will have to incorporate these types of issues, and the development of techniques that can achieve this is the goal of this project. The project's activities will involve algorithm development in the following areas related to the support of mobile computation offloading: (i) energy aware cloud server scheduling; (ii) dynamic network resource allocation; and, (iii) network and cloud server placement and provisioning.The outcomes of this work will extend the operating conditions over which computation offloading will be feasible, permitting its deployment over a much wider range of situations than would otherwise be possible. These results will be of major interest to Canadian network operators, service providers and users.