Tupe, Vaibhav Ramesh (2024) Optimizing Deep Packet Inspection for Securing Remote Work Communication Using Machine Learning: Addressing Performance & Privacy Concerns. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (2MB) | Preview |
Abstract
The exponential rise in remote work in recent years has transformed organizational work fundamentally but it also created an escalating need for the cybersecurity need against the complexity and volume of such cybersecurity threats. Traditional network security techniques like Deep Packet Inspection (DPI) and others face many challenges in actually analyzing the encrypted traffic because of their inherent need for the attack signature patterns thereby reducing their detection against complex and covert cyber threats. This research study focuses on such limitations & challenges by the integration of traditional security measures with advanced Machine Learning (ML) techniques such that real-time traffic classification and threat detection is made possible in remote work environments as well as enterprise environments. This research study uses the comprehensive open-source dataset CICIDS 2017/2018 and various famous ML models including Random Forest, Support Vector Machine (SVM), and XGBoost. Using this dataset and these ML models will accurately classify major network traffic as benign or malicious based on the encrypted data and only using protocol-specific metadata which is extracted using the nDPI library. In the networking world there are many strict privacy standards which must be met and as such this research uses the Encrypted Traffic Analysis (ETA) to ensure user data privacy. This research study utilizes Docker for simulating remote work network environments and Elastic Stack for real-time logging and visualization. Empirical results showed that this proposed DPI-ML framework has performed with exceptional classification accuracy and has also maintained low latency and better throughput while not compromising on the user data privacy. This study was also tested in comparative benchmarking against the existing DPI and ML-based solutions and the results highlight better performance as it not only advances the abilities of traditional DPI but also provides a scalable solution tailored to the dynamic security needs of any remote work environments.
Item Type: | Thesis (Masters) |
---|---|
Supervisors: | Name Email Heffernan, Niall UNSPECIFIED |
Uncontrolled Keywords: | Remote Work; Deep Packet Inspection; Machine Learning; Network Security; Encrypted Traffic Analysis; Real-Time Classification; Privacy Preservation |
Subjects: | Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning |
Divisions: | School of Computing > Master of Science in Cyber Security |
Depositing User: | Ciara O'Brien |
Date Deposited: | 28 Jul 2025 14:38 |
Last Modified: | 28 Jul 2025 14:38 |
URI: | https://norma.ncirl.ie/id/eprint/8270 |
Actions (login required)
![]() |
View Item |