Link prediction for X2 handover interface in LTE networks: Predicting the network topology of X2 LTE interface architecture based on real-time performance metrics data
Blazhko, Veronika (2023)
2023
Master's Programme in Information Technology
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2023-05-16
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202304254433
https://urn.fi/URN:NBN:fi:tuni-202304254433
Tiivistelmä
Predicting the links in a network, based on the node features, has been a prominent problem of the network science of the past few years. The fast changing industry implements more and more complex systems, which creates a demand for network analysis solutions. One of the biggest industries built upon the complex networks is the telecommunications industry, which is working towards increasing the connection speed and decreasing the latency. With the LTE standard introduced, a novel technique of distributing the users has been implemented, that is called 'handing over' or simply 'handover'. Handover technology heavily relies on the topology of the nodes communicating with each other.
In this thesis we propose an effective method of estimating links between the nodes of the network, based on their real-time performance management metrics. Faulty handover interface topology in an LTE network can lead to decreased network performance, inefficient resource allocation and loss of data, ultimately causing a negative impact on user experience. The research is important for such fields as Configuration Management and Performance Management in the telecommunications domain.
The methodology includes Graph Neural Networks (GNNs), which have proven to be useful for graph analytical tasks, such as link prediction and node classification. In this work, we propose a node feature generation method, based on the seasonality analysis and Discrete Fourier Transform (DFT) of the time series signal of each node, combining analytical and machine learning methods for predicting the links in the network. Additionally, in comparison to the latest methods, classic methods will be used as a baseline algorithm.
As a result, we expect the handover interface topology to be recovered and enhanced, user equipment handover accelerated, signal latency is decreased and various business solutions enabled.
In this thesis we propose an effective method of estimating links between the nodes of the network, based on their real-time performance management metrics. Faulty handover interface topology in an LTE network can lead to decreased network performance, inefficient resource allocation and loss of data, ultimately causing a negative impact on user experience. The research is important for such fields as Configuration Management and Performance Management in the telecommunications domain.
The methodology includes Graph Neural Networks (GNNs), which have proven to be useful for graph analytical tasks, such as link prediction and node classification. In this work, we propose a node feature generation method, based on the seasonality analysis and Discrete Fourier Transform (DFT) of the time series signal of each node, combining analytical and machine learning methods for predicting the links in the network. Additionally, in comparison to the latest methods, classic methods will be used as a baseline algorithm.
As a result, we expect the handover interface topology to be recovered and enhanced, user equipment handover accelerated, signal latency is decreased and various business solutions enabled.