Energy Fair Cloud Server Scheduling in Mobile Computation Offloading

Abstract

This thesis considers the issue of energy fairness in mobile computation offloading. In computation offloading, mobile users reduce their energy consumption by executing jobs on a remote cloud server, rather than processing the jobs locally. This can result in energy unfairness however, when the processed jobs are subject to hard deadline constraints.

In this thesis we consider how energy unfairness can be compensated for, by smart scheduling at the cloud server. We first derive an optimum offline scheduler using an integer linear program (ILP) that uses a min-max energy objective and preemptive cloud server scheduling. We then introduce several online scheduling algorithms. The first one is referred to as First-Generated-First-Scheduled (FGFS), where jobs that are generated earlier are given cloud server priority. A modified version, referred to as γ-Ratio Accepted FGFS (γ-FGFS) is proposed, where the acceptance of job submission is subject to an energy threshold test. A version of this algorithm, γ-Ratio Accepted Earliest Deadline First (γ-EDF), is considered that uses earliest deadline first (EDF) scheduling to test for job feasibility.

We also include comparisons using an analytic model that shows the performance possible when the system uses optimum open loop job submission (OLJS) with first-come-first-served (FCFS) cloud server scheduling. Various performance results are presented that show the improvements in energy fairness possible with the proposed schedulers.