Efficient Real-Time Task-Based Scheduling Algorithms for IoT-Fog-Cloud Architecture

Abstract

Integrating the Internet of things (IoT), fog computing, and cloud computing (called IoT-fog-cloud) brings a new direction to solve the problems related to real-life applications, such as self-driving cars, health care, autopilot aircraft, traffic management, fire detection and fighting, and many more. Such applications and their associated tasks are time-sensitive and deadline-oriented. However, executing these tasks in the cloud processors, irrespective of the requirements, cause enormous delay and performance degradation. Therefore, these tasks are categorized into hard, firm and soft; and executed on IoT devices or embedded processors, fog processors and cloud processors, respectively, based on their requirements. But, the execution order of these tasks on different processors is challenging and not emphasized in the literature. Recently, a variant of the earliest deadline first (EDF) has been used to determine the execution order. In this paper, we present three different real-time task scheduling algorithms, namely first come, first served (FCFS), EDF shortest job first (EDF-SJF) and FCFS-SJF for IoT-fog-cloud architecture to determine an efficient ordering of the real-time tasks. These algorithms find the assignment test to allot each task to one of the processors. Then each processor uses min-heap to select a task from its queue in order to execute the same. These algorithms are tested using 100 to 1000 tasks (1000 to 10000 tasks) in a uniprocessor (multi-processor) environment and compared using three performance metrics: success ratio, response time and throughput. We found that FCFS-SJF performs better in most cases than other algorithms.