Enhancing IoT Data-Stream Processing with a Lean Serverless Cloud Architecture

Gudduri, Geethanjali (2025) Enhancing IoT Data-Stream Processing with a Lean Serverless Cloud Architecture. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (917kB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

A modern smart-city system depends on fast data flows of sensor telemetry to support pollution warnings, real-time traffic management and on-demand public services, but the event-processing pipelines of older centralized systems cannot provide sub-second response without high costs for unused resources. The research question behind this study is whether a completely serverless system could fill this gap of latency. A prototype of Amazon Kinesis, AWS Lambda and S3, provisioned using AWS CDK along with CloudWatch, X-Ray and Lambda Insights, shows that sub-second response times are possible. The Kaggle Air Quality in India corpus (approximately 8 million rows) was replayed at throughputs ranging from 25 1 250 messages per second (1 × 50 × baseline). Provisioned concurrency, shard auto-scaling, and 512 MiB memory scaling reduced cold-start InitDuration of 650 ms to less than 10 ms, resulting in a median end-to-end latency of 163 ms and 99 th-percentile of 246 ms even at peak load with no more than 100 ms Kinesis iterator age. The cost analysis showed that pre-warming was cheaper than straight on-demand execution above 400 requests per second, thus verifying the proportional-billing assumption of Function-as-a-Service.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Arun, Shreyas Setlur UNSPECIFIED
Uncontrolled Keywords:	Serverless computing; Function-as-a-Service; IoT data streams; cold-start mitigation; smart-city latency; AWS Lambda; Amazon Kinesis; provisioned concurrency
Subjects:	T Technology > T Technology (General) > Information Technology > Cloud 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 09:48
Last Modified:	26 Mar 2026 09:48
URI:	https://norma.ncirl.ie/id/eprint/9217

Actions (login required)

View Item