Mobile Device Computation Offloading over SocialVPNs

Project Details

Project Lead
Renato Figueiredo 
Project Manager
Renato Figueiredo 
Project Members
Pierre St Juste, Heungsik Eom, Xiaodong Zhang, Kyuho Jeong  
Institution
University of Florida, Electrical and Computer Engineering  
Discipline
Computer Science (401) 

Abstract

The project will advance the fundamental understanding of multiple technical issues faced in enabling flexible end-to-end management of cloud-supported mobile applications. Methods are needed to (re)implement applications so that their software architectures can reconfigure in response to changes in requirements as the environment changes, and to allow for runtime re-factoring of components for execution locally or remotely. Virtual networking techniques are needed to enable seamless off-loading of computation from a device to remote resources possibly provided by multiple clouds. In addition, both software adaptation and network configuration must be done on demand. The FutureGrid project entails research on virtual networking (ad-hoc peer-to-peer networks, social private networks) connecting mobile devices to cloud computing systems for computational offloading.

Intellectual Merit

The proposed work aims to advance the fundamental understanding of multiple technical issues faced in enabling flexible end-to-end management of cloud-supported mobile applications. Methods are needed to (re)implement applications so that their software architectures can reconfigure in response to changes in requirements as the environment changes, and to allow for runtime re-factoring of components for execution locally or remotely. Virtual networking techniques are needed to enable seamless off-loading of computation from a device to remote resources possibly provided by multiple clouds. In addition, both software adaptation and network configuration must be done on-demand. The determination of when and how to adapt must be model-based, i.e. done on the basis of accurate models of software architectures, users, applications and the environment, and monitored system state and desired behaviors; the complexity and real-time nature of this process needs user-centered model-based autonomic solutions. They include the user in the adaptation loop so to optimize for tradeoffs between user satisfaction and other criteria, such as power in the client and server. Interoperability and portability considerations call for the use of new or improved virtualization approaches that are generally applicable to decouple application deployments from physical systems.

Broader Impacts

The project entails research on software engineering (e.g., mapping requirements to software architectures, software architecture models and their adaptation, software re-factoring, adaptable/autonomic middleware, etc), networking (software-defined networking, ad-hoc peer-to-peer networks, social private networks, etc), computer systems (virtualization, distributed computing, etc). The work will also provide techniques and middleware for distributed cyberinfrastructure supportive of mobile-hosted apps for science and education, in particular social virtual private networks connecting mobile devices to cloud infrastructures.

Scale of Use

Usage will be bursty, typically with a few VMs (1-10), occasionally with scalability tests may want to scale up to tens of VMs.