Quasi-opposition Remora Optimizer based Nelder–Mead algorithm for tasks scheduling in cloud

• Published in: Cluster computing 2025-02, Vol.28 (1), p.57, Article 57
• Main Authors: Talha, Adnane, Bouayad, Anas
• Format: Article
• Language: English
• Subjects: Algorithms; Cloud computing; Computer Communication Networks; Computer Science; Critical path; Data encryption; Exploitation; Heuristic; Methods; Operating Systems; Optimization techniques; Particle swarm optimization; Performance evaluation; Processor Architectures; Resource scheduling; Resource utilization; Scheduling; Search methods; Task scheduling; Workflow
• ISSN: 1386-7857; 1573-7543
• DOI: 10.1007/s10586-024-04689-9

The extensive use of cloud computing in several fields creates a number of problems, including resources scheduling, loads balancing, managing power usage, and maintaining security. A suitable scheduling algorithm is required to distribute tasks among the accessible resources in a way that maintains system balance and expedites user task responses in order to achieve the best performance. The three primary quality of service (QoS) criteria: execution time, execution cost, and resource utilization were the motivations behind academics' efforts to improve workflow scheduling (WFS) optimization algorithms. In this study, we present an innovative hybrid method for workflow scheduling. In the proposed RONM, we enhanced Remora Optimization Algorithm search ability using the Nelder–Mead method (NM) to address the local optimum problems by improving the location of potential solutions for weak individuals and improves the capacity of the algorithm to evolve accurately, extending the dimensionality of the optimization problem and speeding convergence to the best solution. Additionally, we use the Quasi-opposition based learning to smart initialize the population and increase the individual’s diversity. Using WorkflowSim, we undertake comprehensive tests to objectively evaluate the performance of our strategy. Our proposed method can improve the overall system performance in terms of quality of service under a wide variety of workload characteristics. According to experimental test based on wide range of real workflows, our approach managed effectively the task scheduling issue when compared to the other comparative approaches like (The Remora strategy (ROA), The Arithmetic Optimization strategy (AOA), The particle swarm method PSO, The Aquila Optimization technique (AO) and The Reptile Search strategy (RSA)). The diversity and validity of RONM solutions are 20.5% and 28.1% better than those of the others solutions.