Analisis Cloud Task Scheduling Menggunakan Multi-Objective Improved Cuckoo Search Algorithm (MOICS) dengan Opposition Based Learning (OBL) pada Cloud Environment

Yusran, Arfan (2025) Analisis Cloud Task Scheduling Menggunakan Multi-Objective Improved Cuckoo Search Algorithm (MOICS) dengan Opposition Based Learning (OBL) pada Cloud Environment. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027211017-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Komputasi awan telah menjadi solusi penting dalam pengembangan layanan berbasis web, di mana peningkatan jumlah pengguna internet menuntut strategi penjadwalan tugas yang efektif untuk optimasi kinerja, efisiensi, dan biaya. Penelitian ini menganalisis implementasi dan kinerja algoritma Multi-objective Improved Cuckoo Search (MOICS) yang diintegrasikan dengan Opposition Based Learning (OBL) untuk penjadwalan tugas di lingkungan komputasi awan. Algoritma MOICS-OBL dibandingkan dengan MOICS standar, Cuckoo Search (CS), dan Particle Swarm Optimization (PSO) menggunakan simulator CloudSim pada dataset SDSC Simple Random, Stratified Random, dan LTL, serta diimplementasikan pada lingkungan nyata berskala kecil. Hasil penelitian menunjukkan bahwa MOICS-OBL secara signifikan meningkatkan performa dibandingkan algoritma pembanding. Secara spesifik, MOICS-OBL unggul dalam parameter makespan dengan peningkatan hingga 37% lebih baik dari CS pada dataset LTL dan 13% lebih baik dari CS/PSO pada Implementasi Nyata. Dalam hal total biaya, MOICS-OBL menunjukkan efisiensi tinggi, yakni 12-15% lebih hemat dibandingkan PSO pada dataset Simple Random dan LTL, serta lebih dari 13% lebih hemat dari semua pembanding pada Implementasi Nyata. Peningkatan juga terlihat pada Average Waiting Time dan Throughput, di mana MOICS-OBL mencapai Throughput hingga 60% lebih tinggi dari CS pada dataset LTL. Selain itu, MOICS-OBL meningkatkan utilisasi sumber daya hingga 29% lebih baik dari semua pembanding pada dataset Simple Random dan mengurangi konsumsi energi hingga 47% lebih hemat dari CS pada dataset LTL. Meskipun PSO terkadang masih unggul dalam beberapa metrik performa pada dataset simulasi yang lebih kompleks, MOICS-OBL menawarkan keseimbangan optimal antara objektif makespan dan biaya. Hasil ini menunjukkan bahwa pendekatan OBL dalam MOICS sangat efektif dalam meningkatkan kinerja penjadwalan tugas, terutama pada kondisi dengan task length rendah dan lingkungan nyata.
====================================================================================================================================
Cloud computing has become an important solution in the development of web-based services,b where the increasing number of internet users demands effective task scheduling strategies for performance optimization, efficiency, and cost. This study analyzes the implementation and performance of the Multi-objective Improved Cuckoo Search (MOICS) algorithm integrated with Opposition Based Learning (OBL) for task scheduling in a cloud computing environment. The MOICS-OBL algorithm is compared with the standard MOICS, Cuckoo Search (CS), and Particle Swarm Optimization (PSO) using the CloudSim simulator on the SDSC, Simple Random, Stratified Random, and LTL datasets, as well as implemented in a small-scale real-world environment. The results show that MOICS-OBL significantly improves performance compared to the comparison algorithms. Specifically, MOICS-OBL excels in the makespan parameter with improvements of up to 37% better than CS on the LTL dataset and 13% better than CS/PSO in the Real-World Implementation. In terms of total cost, MOICS-OBL demonstrates high efficiency, saving 12–15% compared to PSO on the Simple Random and LTL datasets, and over 13% compared to all comparators in the Real-world Implementation. Improvements are also seen in average waiting time and throughput, where MOICS-OBL achieves throughput up to 60% higher than CS on the LTL dataset. Additionally, MOICS-OBL improves resource utilization by up to 29% better than all comparators on the Simple Random dataset and reduces energy consumption by up to 47% more efficiently than CS on the LTL dataset. Although PSO occasionally still outperforms in some performance metrics on more complex simulation datasets, MOICS-OBL offers an optimal balance between makespan and cost objectives. These results demonstrate that the OBL approach in MOICS is highly effective in improving task scheduling performance, particularly under conditions with low task lengths and real-world environments.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Efisiensi, Komputasi Awan, Multi-objective Improved Cuckoo Optimization Algorithm (MOICS), Penjadwalan Tugas, Penghematan Biaya, Cloud Computing, Cost Savings, Efficiency, Task scheduling
Subjects:	Q Science > QA Mathematics > QA76.585 Cloud computing. Mobile computing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Yusran Arfan
Date Deposited:	04 Jul 2025 07:54
Last Modified:	04 Jul 2025 07:54
URI:	http://repository.its.ac.id/id/eprint/119377

Actions (login required)

View Item