Dynamic software deployment between clouds and mobile devices

1. Background

Mobile cloud computing (mCC) is a blend of three foundations, namely cloud computing, mobile computing, and networking enabling more powerful mobile applications by off-loading data and processing of the mobile application to the cloud..

Currently, most mobile applications are being provided directly from online application stores for immediate download and installation in supported smartphones. While smartphones steadily become more powerful to improve the user-experience and provide faster processing of larger and more resource intensive handset-centric applications, in mCC both the data processing and storage is primarily being transferred from the handset itself to the cloud provider, i.e. the handset will be used as a network device to display apps in a mobile browser or through a thin-client interface. Still, many mobile applications like online gaming and augmented reality that require high-processing capacity and minimum network latency will most probably continue to be processed locally on powerful smartphones and mobile tablets while at the same time interacting with services in the cloud. However, mCC poses challenges due to the intrinsic nature and constraints of wireless networks and devices (network latency, limited and dynamically varying bandwidth, battery life).  Thus dynamic solutions wrt use of cloud services (software and platform) by mobile applications are needed to improve the performance and user experience of applications and/or battery life of devices.  In particular there is a need to extend the current state of the art in self-adaptive mobile applications in order to leverage the potential of cloud computing such that an adaptive application on a mobile device

1. may be reconfigured on-the-fly in order to bind to and use a service provided in the cloud if this service provides a higher utility to the application user than what has been in use before, and
2. may “outsource” compute intensive application or middleware components to the cloud, for example in order to gain performance, save battery resources, or obtain more storage capacity.

This master thesis proposal will focus on the latter.

2. Master thesis proposal

Dynamic software deployment between clouds and mobile devices

The thesis will be concerned with finding efficient solutions for how to dynamically adapt the mobile applications and middleware partition decisions between a mobile device and one or more cloud platforms to dynamically update the distributed deployment of the running software in response to context changes such as changes in bandwidth, CPU utlization on the device, data loads and discovered (nearby) cloud platform services (cf CloneCloud[1]).

In this project it is proposed to leverage the MUSIC adaptation middleware (http://ist-music.berlios.de) to the domain of mobile cloud conputing. This middleware originally targets mobile application in ubiquitous computing environments. However, MUSIC is designed as a framework allowing the customisation of the middleware to application or application domain specific needs and to different execution environments, both hardware and software wise. The main task of this thesis will be to figure out how MUSIC need to be customized in order to support dynamic software deployment between  clouds and mobile devices.

The thesis would include the following tasks.

• Study of the MUSIC middleware to learn how MUSIC can be customized.
• Study mobile-cloud computing to identify requirements to dynamic software deployment between  clouds and mobile devices  and limitations of  state-of-the-art.
• Based on the above study identify what need to be customized in MUSIC to support dynamic software deployment as above.
• Address one or more of the customization needs in detail (design, implementation, evaluation).

The thesis can probably be done by two students in collaboration, sharing the work on the MUSIC middleware, requirements and overall customization study, but addressing separate customizations needs in detail, separately.

3. International collaboration

The master thesis will be done in a collaboration with former MUSIC project partners, including University of Kassel, Germany