Cooperative Game Theoretical Techniques for Energy-Aware Task Scheduling in Cloud Computing

Abstract

Energy efficiency is at the forefront of challenges faced by the cloud providers. Lowering the power usage helps providers achieve the Energy Star rating, which is important from marketing and environmental perspectives. Developed in this paper is an energy-aware scheduling algorithm for distributed heterogeneous machines in a cloud environment. With methodological modeling, the Energy Aware Task Allocation (EATA) problem is converted into a bargaining game. A scheduling algorithm is created and strongly proved, axiom by axiom, to generate a Nash Bargaining Solution (NBS), with faster turnaround time and fewer relaxed system assumptions compared to the previous work. A complementary business revenue equation incorporating Dynamic Voltage Frequency Scaling (DVFS) is utilized that rewards energy efficient users through cost savings. The contribution of the paper is the novel methods of designing gaming mechanisms for the EATA problem. In addition, these methods are evaluated using the proposed business revenue model, as is confirmed by analysis and experimental framework.