EMUNE: a bulk transfer service architecture for multi-interfaced mobile devices

Abstract

Today's mobile devices increasingly contain multiple radios, enabling users on the move to take advantage of a heterogeneous wireless network environment. In addition, many bandwidth intensive services, such as podcasts or software updates are highly delay tolerant and memory in mobile devices is effectively unlimited. To take advantage of these technology trends, we propose EMUNE, a new transfer service architecture. EMUNE supports opportunistic bulk transfers in high bandwidth networks with intermittent availability. In addition, EMUNE can also satisfy a minimum data transfer rate, and dynamically adapt to application requirements and user preferences for quality and monetary/power costs.

Our proposed architecture consists of an API, a prediction engine, a scheduling engine and a transport service. The API specifies what the transfer service offers, and how the service can be used by the applications. The prediction engine infers future network availability and their characteristics such as the available bandwidth for a time duration ahead. The scheduling engine takes the output of the prediction engine as well as the power and monetary costs of networks, application requirements and user preferences into account. It then determines which interface to use, when and for how long for all outstanding data transfer requests within that duration. The transport service accordingly executes the inferred data transfer schedule by assigning applications' data flows to the corresponding access networks. The architecture repeats this process for successive time durations, as long as there is data available to be transferred.

In addition to presenting EMUNE, we provide a thorough evaluation of the architecture and its components using data of many real users, to show the effectiveness of EMUNE in improving the utilization of multiple network interfaces of mobile devices in a heterogeneous networking environment.