Deep learning and optimization-based task scheduling algorithms for fog-cloud computing environment

Article type: Research Article

Authors: Ahamed, Ayoobkhan Mohamed Uvaze^{a; *} | Joel Devadass Daniel, D.J.^b | Seenivasan, D.^c | Rukumani Khandhan, C.^d | Radhakrishnan, S.^e | Daya Sagar, K.V.^f | Bhardwaj, Vivek^g | Nishant, Neerav^h

Affiliations: [a] Software Engineering, New Uzbekistan University, Tashkent, Uzbekistan | [b] Department of ECE, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology, Chennai, Tamilnadu, India | [c] Computer Science and Business Systems, M. Kumarasamy College of Engineering, Karur, Tamilnadu, India | [d] Kongu Engineering College, Perundurai, Erode, Tamilnadu, India | [e] Department of CSE-AI, KKR & KSR Institute of Technology & Sciences, Guntur, Andhra Pradesh, India | [f] Department of Electronics and Computer Engineering, Koneru Lakshmaiah Education Foundation, Vaddeswaram, Guntur, Andhra pradesh, India | [g] Manipal University Jaipur, Jaipur, Rajasthan, India | [h] Department of Computer Science and Engineering, School of Engineering, Babu Banarasi Das University, Lucknow, Uttar Pradesh, India

Correspondence: [*] Corresponding author. Ayoobkhan Mohamed Uvaze Ahamed, Software Engineering, New Uzbekistan University, Tashkent, Uzbekistan. E-mail: [email protected].

Abstract: Time-sensitive programs that are linked to smart services, such as smart healthcare as well as smart cities, are supported in large part by the fog computing domain. Due to the increased speed limitation of the cloud, Cloud Computing (CC) is a competent platform for fog in data processing, but it is unable to meet the demands of time-sensitive programs. The procedure of resource provisioning, as well as allocation in either a fog-cloud structure, takes into account dynamic changes in user requirements, and resources with limited access in fog devices are more difficult to manage. Due to the continual changes in user requirement factors, the deadline represents the biggest obstacle in the fog computing structure. Hence the objective is to minimize the total cost involved in scheduling by maximizing resource utilization. For dynamic scheduling in the fog-cloud computing model, the efficiency of hybridization of the Grey Wolf Optimizer (GWO) and Lion Algorithm (LA) is developed in this study. In terms of energy costs, processing costs, and communication costs, the created GWOMLA-based Deep Belief Network (DBN) performed better and outruns the other traditional models.

Keywords: Fog-cloud computing environment, deep learning, deep belief network (DBN), lion algorithm (LA), grey wolf optimizer (GWO).

DOI: 10.3233/JIFS-234030

Journal: Journal of Intelligent & Fuzzy Systems, vol. Pre-press, no. Pre-press, pp. 1-14, 2023