Improving Elastic FL Efficiency in Serverless Environments using Dynamic Resource Allocation and Intelligent Client Selection

Inbaraj, Silver Stalan (2025) Improving Elastic FL Efficiency in Serverless Environments using Dynamic Resource Allocation and Intelligent Client Selection. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (920kB) | Preview
Preview	PDF (Configuration Manual) Download (293kB) | Preview

Abstract

Federated Learning (FL) has emerged as a pioneering form of distributed machine learning methodology that allows for the collaborative training of models over widely different devices with privacy of the data kept in mind. The traditional FL on top of public clouds is notably problematic and includes resource underutilization, the existence of the straggler effects, and poor client selection strategies especially in heterogynous settings that comprise computing capabilities and data distributions. The proposed research gives an original concept of the elastic FL framework that combines AWS autoscaling benefits with the statistical analysis algorithm to implement dynamic resources allocation scheme within serverless computing frameworks. The framework uses asynchronous aggregation strategy with a score-based client selection scheme that analyses the capacity and capability of the client, the nature of the data and the previous performance evaluation of the client. The Resource Optimizer uses Median Absolute Deviation (MAD) statistical analysis and policy-based scaling reasoning to optimize the use of resources, which does not require detailed machine learning predictions. Thorough comparison with the Vertical Pod Autoscaler in Kubernetes obtained a smaller training time (30%), saved cost (40%), and improved resource usage (21%) compared to the current best practices, making the case that statistical methods can be effective to optimize serverless FL.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Emani, Sai UNSPECIFIED
Subjects:	T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > QA Mathematics > Electronic computers. Computer science > Computer Systems > Computers > Electronic data processing > Electronic data processing--Distributed processing > Edge computing T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science > Computer Systems > Computers > Electronic data processing > Electronic data processing--Distributed processing > Edge computing T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Telecommunications > Computer networks > Internet of things Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	26 Mar 2026 10:06
Last Modified:	26 Mar 2026 14:35
URI:	https://norma.ncirl.ie/id/eprint/9220

Actions (login required)

View Item