Fog-cloud scheduling simulator for reinforcement learning algorithms

• Main Authors: Al-Hashimi, Mustafa Ahmed Adnan, Rahiman, Amir Rizaan, Muhammed, Abdullah, Hamid, Nor Asilah Wati
• Format: Article
• Published: Springer 2023
• Online Access: http://psasir.upm.edu.my/id/eprint/108029/; https://link.springer.com/article/10.1007/s41870-023-01479-1?error=cookies_not_supported&code=90ee88c7-2096-44ad-a4cc-ab5fca21f9a6

Fog computing is a popular choice for Internet of Things (IoT) applications, such as electricity, health, transportation, smart cities, security, and more. Due to its decentralized architecture, fog computing offers low latency processing and ensures the preservation of information between the source and the cloud resources. Additionally, it can be integrated with the cloud to provide satisfactory and efficient service simultaneously. However, the main challenge with fog computing is that the edge nodes, called fog devices, have limited processing capabilities and storage for dynamic high-level operations. Therefore, supplying optimized scheduling algorithms to provide satisfactory quality service for the node’s task execution and processing becomes demanding. Most existing simulators are built based on simplified situations, which causes degradation in performance when realistic scenarios are considered. This study presents a developed simulator that captures all mentioned realistic scenarios by providing the feature of integrability with the reinforcement learning (RL) algorithm. Furthermore, three validation steps have been used to measure the simulator’s effectiveness: real-time visualization, intense task arrival, and preservation test have been used, and the results proved the simulator suitable for dealing with realistic situations.