Bilgilendirme: Kurulum ve veri kapsamındaki çalışmalar devam etmektedir. Göstereceğiniz anlayış için teşekkür ederiz.

Browsing by Author "Osama, Muhammad"

Filter results by typing the first few letters
Now showing 1 - 1 of 1
  • Thumbnail Image
    Article
    AGENT: An Elitism-Guided Evolutionary Framework for Enhanced Task Allocation Performance in Heterogeneous Cloud Systems
    (Elsevier, 2026) Osama, Muhammad; Riaz, Muhammad Bilal; Shahid, Muhammad Farrukh; Qadri, Syed Shah Sultan Mohiuddin
    Cloud Computing (CC) delivers on-demand services through Infrastructure as a Service (IaaS), Platform as a Service (PaaS), and Software as a Service (SaaS) models. This study specializes in the approach to IaaS task scheduling in the heterogeneous data centers, where resource allocation is a critical issue to ensure that makespan is minimized. The NP-complete nature of such a scheduling problem requires sophisticated meta-heuristic solutions as the use of cloud workloads increases exponentially. Although Genetic Algorithms (GA) have received extensive use, available adaptive variants usually vary parameters based on aggregate populations statistics without individual solution tracking and elitism is not commonly implemented with adaptive mechanisms in a size-conserving fashion. This paper presents the Adaptive Genetic Algorithm with Elitism and Nonlinear Tuning (AGENT) that combines three new innovations: (1) size-preserving elitism that guarantees a monotonic improvement without growing the population, (2) feedback-based nonlinear parameter adaptation which is controlled by explicit success/failure counters as indicators of evolutionary progress of a population, unlike fitness-proportional or population-statistics-based methods, and (3) a multi-task-per-VM allocation model that captures real cloud elasticity. Experimental validation of CloudSim Plus simulation with Amazon EC2 VM setups showed makespan improvements of 3.14%-28.89% than baseline algorithms (HAGA, AIGA, SGA, Max-Min, Min-Min) with synthetic workloads. Scalability was tested on workloads of various sizes and was found to perform well with near-optimal results. Reduced makespan is associated with shorter VM operating time, which implies that energy efficiency may be improved and therefore, it should be investigated in the future by taking direct measurements.