Federated Learning for Cyberattack Detection in Decentralized Networks

Tran, Viet Khoa

Federated Learning for Cyberattack Detection in Decentralized Networks

Tran, Viet Khoa

Permalink

Publication Type:: Thesis
Issue Date:: 2024

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download thesisAdobe PDF (26.88 MB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Tran, Viet Khoa
dc.date.accessioned	2024-06-19T00:50:05Z
dc.date.available	2024-06-19T00:50:05Z
dc.date.issued	2024
dc.identifier.uri	http://hdl.handle.net/10453/179560
dc.description	University of Technology Sydney. Faculty of Engineering and Information Technology.	en_US.UTF-8
dc.description.abstract	Recently, the rapid development of various technologies, such as blockchain and Internet-of-Things (IoT), has enabled numerous applications to become integral to many aspects of our daily lives. However, this leads to a massive amount of data and raises serious security concerns. Machine Learning (ML), especially Deep Learning (DL), has been widely used in cyberattack detection to detect cyberattacks in emerging networks. Nevertheless, DL-based cyberattack detection systems usually require a huge amount of data from users/systems. This threatens user privacy because sensitive data may be sent over the network to a centralized server for processing. Additionally, transmitting such a large amount of data imposes communication overhead on the network. This thesis aims to develop ML frameworks that can efficiently detect cyberattacks/intrusions in decentralized networks, such as IoT and blockchain networks, without exposing their local data over the network. In the first study, we develop a collaborative learning framework that enables a target network with unlabeled data to learn "knowledge" from a source network with abundant labeled data in IoT networks. Our proposed framework can exchange the learned "knowledge" among various DL models, even when their datasets have different features. The experiments showed that our proposed framework could achieve higher accuracy than state-of-the-art DL-based approaches. In the second study, we develop a cyberattack detection framework to detect cyberattacks in the network traffic of blockchain networks. Specifically, we first implement a blockchain network in our laboratory. This blockchain network will serve to generate real traffic data and implement real-time experiments to evaluate the performance of our frameworks. We then propose a collaborative learning model that allows efficient deployment in the blockchain network to detect attacks. Both simulated and real-time experiments have been provided to demonstrate the efficiency of our proposed framework. In the third study, we propose a collaborative learning framework to detect attacks on blockchain transactions and smart contracts. Our framework can classify various types of blockchain attacks, including intricate attacks at the machine code level which typically necessitate significant time and security expertise. Additionally, our framework enables real-time detection of diverse attack types at Ethereum nodes. The simulated and real-time experiments demonstrate the outperformance of our proposed framework compared to conventional DL models. All the results above have demonstrated that our proposed federated learning-based frameworks can efficiently be deployed in decentralized networks to detect cyberattacks with high accuracy.	en_US.UTF-8
dc.format	Thesis (PhD)
dc.language.iso	en_US	en_US.UTF-8
dc.relation	https://opus.lib.uts.edu.au/bitstream/10453/179560/1/thesis.pdf
dc.rights	info:eu-repo/semantics/openAccess
dc.rights	The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.
dc.rights	© 2024 Viet Khoa Tran
dc.rights	au.edu.uts.lib/nph
dc.title	Federated Learning for Cyberattack Detection in Decentralized Networks	en_US.UTF-8
dc.type	Thesis
utslib.copyright.status	open_access	*

Abstract:

Recently, the rapid development of various technologies, such as blockchain and Internet-of-Things (IoT), has enabled numerous applications to become integral to many aspects of our daily lives. However, this leads to a massive amount of data and raises serious security concerns. Machine Learning (ML), especially Deep Learning (DL), has been widely used in cyberattack detection to detect cyberattacks in emerging networks. Nevertheless, DL-based cyberattack detection systems usually require a huge amount of data from users/systems. This threatens user privacy because sensitive data may be sent over the network to a centralized server for processing. Additionally, transmitting such a large amount of data imposes communication overhead on the network. This thesis aims to develop ML frameworks that can efficiently detect cyberattacks/intrusions in decentralized networks, such as IoT and blockchain networks, without exposing their local data over the network. In the first study, we develop a collaborative learning framework that enables a target network with unlabeled data to learn "knowledge" from a source network with abundant labeled data in IoT networks. Our proposed framework can exchange the learned "knowledge" among various DL models, even when their datasets have different features. The experiments showed that our proposed framework could achieve higher accuracy than state-of-the-art DL-based approaches. In the second study, we develop a cyberattack detection framework to detect cyberattacks in the network traffic of blockchain networks. Specifically, we first implement a blockchain network in our laboratory. This blockchain network will serve to generate real traffic data and implement real-time experiments to evaluate the performance of our frameworks. We then propose a collaborative learning model that allows efficient deployment in the blockchain network to detect attacks. Both simulated and real-time experiments have been provided to demonstrate the efficiency of our proposed framework. In the third study, we propose a collaborative learning framework to detect attacks on blockchain transactions and smart contracts. Our framework can classify various types of blockchain attacks, including intricate attacks at the machine code level which typically necessitate significant time and security expertise. Additionally, our framework enables real-time detection of diverse attack types at Ethereum nodes. The simulated and real-time experiments demonstrate the outperformance of our proposed framework compared to conventional DL models. All the results above have demonstrated that our proposed federated learning-based frameworks can efficiently be deployed in decentralized networks to detect cyberattacks with high accuracy.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/179560