Comparison of Edge Computing Scheduling Algorithms

Abstract

Through the advancement of the Internet of Things (IoT), the development of devices for task automation, data extraction, and communication between devices has become increasingly easy. But as a result, tens of zettabytes of data are being generated every year, causing excessive bandwidth consumption as well as slow response times for devices. One of the ways to solve the problem is with the use of Edge Computing networks, such paradigm allows the transfer of the data processing to the edges of the network. Since the Edge is mostly composed of devices of varied and limited computational capacity, a good way to distribute the tasks that must be processed is needed. Therefore, efficient, and well tested, scheduling algorithms are a way to distribute tasks in such a way that the time required to perform them is minimized. This work explores the comparison o three distinct scheduling algorithms in Edge Computing: the Modified Monte Carlo Tree Search; the Improved Binary Grey Wolf Optmizer and the Application-aware Scheduling Algorithm, analyzing their speed and efficiency as an evaluation metric, using the iSPD grid simulator.