Bridging the Gap Between Detection and Enforcement: A Critical Review of Deep Learning-Based Zero-Trust Intrusion Detection Models

Saqib, Muhammad (2025) Bridging the Gap Between Detection and Enforcement: A Critical Review of Deep Learning-Based Zero-Trust Intrusion Detection Models. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (645kB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

The conventional security systems which depend on barriers and walls around the perimeter are usually not effective against the current hackers and insider threats. A solution to this is the integration of an algorithm (Deep Learning (DL)) and an architecture (Zero-Trust Architecture (ZTA)). In the old security systems, it was the fixed rules, a factor that was slow and could not cope with the emerging attacks. In order to repair such issues, this study develops a new model that integrates Deep Learning and Zero Trust. The system automatically detects threats and prevents them within the company networks. The system is based on the state-of-the-art AI techniques namely, convolutional and recurrent neural networks. They were trained with the CICIDS2017 and Edge-IIoTset datasets to identify suspicious behavior in the network. A Python program functions as an engine of a Zero-Trust policy after the AI identifies a problem to govern access, in real-time, as a form of control. To verify the accuracy, speed, and flexibility of this system, we experimented with it and added an automatic trust score to enhance the past experiments. As can be seen, our 1D-CNN model performed quite well with a 96 percent accuracy. It was able to distinguish between malicious traffic and harmless traffic. Most importantly, the system is very fast; a check of each two data packets requires less than 10 milliseconds. This demonstrates that it can be used in real-time at high-speed networks. The model also tested the data of the Industrial IoT (Edge-IIoTset) and could adjust to this new medium. The policy engine was able to make an automatic decision with two options; Block, Flag or allow without the assistance of a human being. This paper provides a viable and repeatable approach to create smart Zero-Trust security. It also demonstrates a scaling of this system in Cloud and IoT. Nevertheless, there are still obstacles, like the necessity to continually update the model and keep security regulations up-to-date.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Sahni, Vikas UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security 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 Cyber Security
Depositing User:	Ciara O'Brien
Date Deposited:	04 Sep 2026 10:11
Last Modified:	04 Sep 2026 10:11
URI:	https://norma.ncirl.ie/id/eprint/9826

Actions (login required)

View Item