Multi-criteria aware task scheduling algorithm based on PSO for fog computing

Abstract

In this paper, we propose a task scheduling method for applications made of independent tasks executed on a fog computing network. Fog computing is a computing network architecture extending the cloud computing model. It adds an intermediate computational layer between IoT devices and the cloud to solve the computational requirements of the current IoT era. This fog layer or fog network uses resources from network infrastructures such as base stations, routers, switches, or servers. These resources, also called fog nodes, have computational capabilities so that they can process tasks, and we consider that they have additional properties such as link characteristics or security threat levels. Task scheduling consists of the attribution of the tasks of applications coming from IoT devices to the resources of the fog network, according to a predefined objective. However, finding optimal task scheduling solutions is a np-hard problem, so that is the reason we use a heuristic approach in our research. The paper presents a task scheduling method for fog computing based on particle swarm optimization (PSO). This scheduling method considers several criteria defined as objectives. These criteria are the makespan, the cost of processing, the cost of using network links, and the security cost involved in the processing of the application. Then, we evaluate the performance of the provided scheduling method in a simulated fog network environment and evaluate the impact of a specific criterion.