Multi-Objective Optimization Approach for Task Scheduling in Fog Computing

The Fog computing paradigm allow applications to be processed at the edge of a network. This paradigm is designed to mitigate high latency and the burden of task requests sent to centralized cloud servers by end devices. Fog computing permits different portions of applications to be scheduled to Fog nodes available at the edge. These Fog nodes offer cloud processing services and have appeared as a feasible technique for real time applications. However, scheduling a task among available Fog nodes must be effective, meaning it must not over consume available resources because of limited resources at the edge. Consuming extra amount of energy than available on the Fog nodes can lead to network breakdown or application failure which is not acceptable for real-time applications. Therefore, to address this challenge, this paper presents an application scheduling technique based on virtualization technology to find an efficient algorithm that can optimize energy consumption and average delay of real-time applications in Fog computing networks. This is achieved by implementing four task scheduling policies in a Fog node scheduler to assess their performance and efficiency. Simulations were conducted using the iFogSim tool and the results demonstrate that the FCFS scheduling policy achieved improvement in energy consumption by 11 %, average task delay 7.78 %, 4.4 % network usage and execution time 15.1 % better than other algorithms.