Real-Time Scheduling on Dynamic Resources in a Fog Computing Environment

摘要

A multitude of computing devices are networked around us but most do not harness their computation capacity to the fullest. With this as the core motivation, Cisco delivered the vision of fog computing to run applications on local nearby devices at network edge instead of the cloud.Fog's close proximity to client devices, dense geographical distribution and support for mobility provides end users with a cheap computational resource available at very fast speeds. We extend this architecture through the inclusion of dynamic resources, each of which consists of the residualcomputing capacity of a particular edge device. Thus, edge devices are allowed to dynamically arrive into the system for an interval known a priori (at the time of their arrival) and lend a fraction of their computing capacities against financial incentives. We present a novel two level frameworkfor the symbiotic sharing and consolidation of edge computing resources targeted towards realizing a real time fog computing environment with high resource utilization. Further, the framework has been extended to imbibe profit awareness by ensuring a minimum guaranteed revenue to the fog serviceprovider. Additionally, inter-device communication delays are controlled by limiting the communication overheads in the system. Our simulation based experiments on diverse datasets have procured promising results.