GreenFog: A framework for sustainable Fog computing

The alarming rate of increase in energy demand and carbon footprint of Fog environments has become a critical issue. It is, therefore, necessary to reduce the percentage of brown energy consumption in these systems and integrate renewable energy use into Fog. Renewables, however, are prone to availability fluctuations due to their variable and intermittent nature. In this paper, we propose a new Fog framework and design various optimization techniques, including linear programming optimization, linear regression estimation, and Multi-Armed Bandit (MAB) learning to optimize renewable energy use in the Fog based on a novel idea of load shaping with adaptive Quality of Service (QoS). The proposed framework, along with the optimization techniques, are tested on a real-world micro data center (Fog environment) powered by solar energy sources connected to multiple IoT devices. The results show that our proposed framework significantly reduces the difference between renewable energy generation and total energy consumption while efficiently adjusting the QoS of applications.