OaaS: offload as a service in fog networks

Abstract

Cloud computing is a mature technology that provides a huge leap in elastic computation, and new development trends are shaped to compliment the cloud computing paradigm. Cisco recently introduced the concept of Fog Computing to enable applications on billions of devices that are already connected and form the Internet of Things at the edge of the network. With the view point that the fog computing paradigm will be the future of computing technology, we look at its strong characteristics and propose a novel approach to enable a new kind of service called Offload As A Service (OaaS). Offload computation has been an active research area for many years and provides the capability to extend mobile resources limitations in terms of CPU, GPU, memory, storage and battery energy. Fog computing paradigm is a good synergy for offload computation technology with its low delay and close proximity features. To realize the enabling of OaaS in a fog computing environment, we propose a novel framework for communication and an offloading mechanism between different layers of the fog infrastructure, using matching algorithm to handle the fair mapping between users and service providers. Simulation results are provided to validate the effectiveness of our proposal. Simulation results of our work have shown great potential and value based on the prototype implementation, especially in the running time of tasks, which was reduced significantly with improvements of up to 37 and $12\%$ observed for photos offloaded to “PC” and “Odroid”, respectively, compared to the local running time on the user device.