Resource Allocation Mechanism for a Fog-Cloud Infrastructure

Abstract

Fog computing brings the cloud close to the users, reducing latency and allowing the deployment of new delay sensitive applications. Fogs and clouds can work cooperatively to improve service delivery to the end users. An essential aspect of a fog-cloud system is the decision-making process on where to allocate resources to run the tasks of an application. This paper introduces a novel mechanism named Gaussian Process Regression for Fog-Cloud Allocation (GPRFCA) for resource allocation in infrastructure composed of cooperative fogs and clouds. The GPRFCA mechanism employs a Gaussian Process Regression to predict future demands in order to avoid blocking of requests, especially delay-sensitive ones. Results show that the GPRFCA mechanism reduces energy consumption, blocking as well as latency.