University of Oulu

Federated learning-based anomaly detection as an enabler for securing network and service management automation in beyond 5G networks

Saved in:
Author: Jayasinghe, Suwani1
Organizations: 1University of Oulu, Faculty of Information Technology and Electrical Engineering, Department of Communications Engineering, Communications Engineering
Format: ebook
Version: published version
Access: open
Online Access: PDF Full Text (PDF, 2.6 MB)
Pages: 73
Persistent link: http://urn.fi/URN:NBN:fi:oulu-202205172146
Language: English
Published: Oulu : S. Jayasinghe, 2022
Publish Date: 2022-05-18
Thesis type: Master's thesis (tech)
Tutor: Ylianttila, Mika
Reviewer: Ylianttila, Mika
Porambage, Pawani
Description:

Abstract

Zero-touch network architecture (ZSM) is proposed to cater to unprecedented performance requirements, including network automation. 5G and beyond networks include exceptional latency, reliability, and bandwidth requirements. As a result, network automation is a necessity. ZSM architecture combines closed-loop mechanisms and artificial intelligence (AI) to meet the network automation requirement. Even though AI is prevalent, privacy concerns and resource limitations are growing concerns. However, techniques such as federated learning (FL) can be applied to address such issues. The proposed solution is a hierarchical anomaly detection mechanism based on the ZSM architecture, divided into domains by considering technical or business features. The network flow is categorized as an anomaly or not, and abnormal flows are removed from both stages. Detectors and aggregation servers are placed inside the network based on their purpose. The proposed detector is simulated with the UNSW-NB15 Dataset. The simulation results show accuracy improvement after the 2nd stage, and the detection accuracy varies with training data composition.

see all

Subjects:
Copyright information: © Suwani Jayasinghe, 2022. Except otherwise noted, the reuse of this document is authorised under a Creative Commons Attribution 4.0 International (CC-BY 4.0) licence (https://creativecommons.org/licenses/by/4.0/). This means that reuse is allowed provided appropriate credit is given and any changes are indicated. For any use or reproduction of elements that are not owned by the author(s), permission may need to be directly from the respective right holders.
  https://creativecommons.org/licenses/by/4.0/