FlexConnect : Mobile Computational Offloading

Abstract

Edge computing is increasing its popularity also in non-computing specific environments. In particular, Makerspaces are facing the issue of providing cognitive-edge to cloud computational power to their users, for running their 3D modeling and optimizations. However, while the edge-to-cloud offloading might be a traditional issue in edge computing systems, some countries require Makerspaces to provide their service only during operational hours, and therefore to switch off their machines, including their servers, during the night. This calls for the need for flexible computational offloading, enabling to checkpoint of current tasks and restoring them during opening hours. This work examines the feasibility of designing and implementing such a system, to facilitate computational offloading with checkpointing to increase the availability of computational power. We propose and prototype the implementation for a system that enables a provider to provide computational power through their edge with a checkpoint and restore feature to prevent loss of progress while the system works on a computation. A client-side application was designed and implemented to enable users to access the computational offloading service from a provider at a given location directly from their smartphone. Results show that it is possible to implement a continuous edge-to-cloud computational platform to enable full checkpoint and restore of computational tasks while providing a good quality of service and using a controlled amount of energy.