Bovenzi, Giampaolo (2022) A hierarchical learning framework for network traffic analysis: design, implementation, and use cases. [Tesi di dottorato]


Download (5MB) | Preview
[error in script] [error in script]
Item Type: Tesi di dottorato
Resource language: English
Title: A hierarchical learning framework for network traffic analysis: design, implementation, and use cases.
Date: 14 March 2022
Number of Pages: 154
Institution: Università degli Studi di Napoli Federico II
Department: Ingegneria Elettrica e delle Tecnologie dell'Informazione
Dottorato: Information technology and electrical engineering
Ciclo di dottorato: 34
Coordinatore del Corso di dottorato:
Pescapè, AntonioUNSPECIFIED
Date: 14 March 2022
Number of Pages: 154
Keywords: network traffic analysis; traffic classification; intrusion detection; hierarchical learning; deep learning; machine learning
Settori scientifico-disciplinari del MIUR: Area 09 - Ingegneria industriale e dell'informazione > ING-INF/05 - Sistemi di elaborazione delle informazioni
Date Deposited: 22 May 2022 21:35
Last Modified: 28 Feb 2024 11:00

Collection description

Network traffic analysis covers the entire set of operations and techniques used to gain knowledge about the status of a network, in order to manage and administer it properly. Therefore, traffic analysis solutions resort to modeling techniques applied to the network traffic with the aim of aiding network operators and internet service providers to achieve a clear snapshot of what is traversing their network. The main challenges we identified in the nowadays Internet traffic are strictly related to the (i) the huge number of Internet enabled devices which generates heterogeneous network traffic; and (ii) the increasing of the generated traffic in terms of traffic volume. Accordingly, the main objective of this doctoral thesis is proposing a Hierarchical Learning Framework for Network Traffic Analysis, in order to enhance the fine-grained network knowledge and the scalability of traffic analysis solutions. This framework enhances traffic analysis exploiting hierarchical dependencies among network traffic classes in order both to improve the fine-grained modeling of network traffic and to enable a modular and scalable learning process, enabling fast retraining. To this extent, the proposal takes advantage of state-of-the-art machine and deep learning solutions, thus fostering designed hierarchical learning approaches and it is evaluated on different types of Internet traffic (viz. three scenarios), such as the traffic generated by privacy-preserving solutions (e.g., VPNs, Anonymity Tools), network attacks or malware (e.g., Scans, Denial of Services, Botnets), mobile applications (e.g., Games, Social Networks, Video on Demand).


Downloads per month over past year

Actions (login required)

View Item View Item