OPTIMIZING CLOUD RESOURCE ALLOCATION FOR IOT SYSTEMS USING ML APPROACHES

K S Saraswathi Devi

OPTIMIZING CLOUD RESOURCE ALLOCATION FOR IOT SYSTEMS USING ML APPROACHES

Authors: K S Saraswathi Devi

Unique Paper ID: 167326
Volume: 4
Issue: 3
PageNo: 294-300

Keywords: Cloud Resource XGBoost Neural Networks Internet of Things Random Forest

Abstract:
This research investigates the use of several machine learning models to optimise cloud resource allocation in Internet of Things applications. Using a collection of resource allocation measurements from many IoT implementations, a thorough analysis was carried out, evaluating models for cost effectiveness, resource utilisation, and forecast accuracy. The findings show that, with a Mean Absolute Error (MAE) of 2.89 and a Root Mean Square Error (RMSE) of 4.98, XGBoost had the best prediction accuracy. The Neural Network came in second, with an MAE of 3.01 and an RMSE of 5.12. Moreover, Random Forest performed admirably, showing an RMSE of 5.34 and an MAE of 3.12. XGBoost and Neural Networks had the greatest average CPU and memory utilisation, at 33.5% and 35.8%, respectively, in terms of resource utilisation. Decision trees demonstrated reduced resource use, with an average CPU utilisation of 28.7% and memory usage of 110 MB, but being significantly less accurate (MAE of 4.56, RMSE of 6.78). Cost analysis indicated that XGBoost incurred the highest total monthly cost at $2700, followed by Neural Networks at $2800. In contrast, Decision Trees proved to be the most cost-effective with a total monthly cost of $2400. The study concludes that while XGBoost and Neural Networks provide superior accuracy, their higher operational costs may not be justified in all scenarios. Decision Trees, though less accurate, present a more cost-effective solution, making them suitable for environments with strict budget constraints.

Download article

email to a friend

Cite This Article

ISSN: 2349-6002
Volume: 4
Issue: 3
PageNo: 294-300

OPTIMIZING CLOUD RESOURCE ALLOCATION FOR IOT SYSTEMS USING ML APPROACHES

Available:https://ijirt.org/Article?manuscript=167326

15 Aug, 2024

SCALABLE IOT DATA MANAGEMENT IN CLOUD ENVIRONMENTS: TECHNIQUES AND CHALLENGES

16 Feb, 2021

Carrier App: Leveraging Machine Learning for Secure Crowdshipping

12 Feb, 2021

SafeLid Vision: A Machine Learning Approach to Automated Helmet Detection

Impact Factor
8.01 (Year 2024)

UGC Approved
Journal no 47859

Join Our IPN

IJIRT Partner Network

Submit your research paper and those of your network (friends, colleagues, or peers) through your IPN account, and receive 800 INR for each paper that gets published.

Join Now

Latest Publication

Recent Conferences

NCSEM 2024

National Conference on Sustainable Engineering and Management - 2024 Last Date: 15th March 2024

Submit inquiry