Microservices-Based Autonomous Anomaly Detection for Mobile Network Observability

Loading...
Thumbnail Image

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Master's thesis

Date

2023-10-09

Department

Major/Subject

Data Science

Mcode

SCI3115

Degree programme

Master's Programme in ICT Innovation

Language

en

Pages

122+7

Series

Abstract

In modern telecommunication networks, network observability entails the use of diverse data sources to understand the state and behavior of the network, and its ability to provide the required service and user experience. Because of the vast amounts of data collection and transmission involved in this process, the network's performance is negatively impacted, and it can become difficult for network operators to identify the occurrence of problematic behavior before it is too late. To enable a more efficient form of data collection and aid in diagnostic operations, this thesis aims to develop an autonomous anomaly detection system for time series data. The system is to be developed as a microservices-based solution, to be integrated with a software-defined networking controller platform developed at \textit{Ericsson}. This thesis describes the extensive experimentation process conducted during the development of this system, including various methods of data processing, time series clustering, and anomaly detection. The resulting system is a highly customizable and scalable product, supported by modern and reliable anomaly detection models. The system is capable of detecting several different kinds of anomalies in an arbitrary number of mobile network monitoring metrics and can be easily configured to fit the specific needs of each customer.

Description

Supervisor

Jung, Alex

Thesis advisor

Bergenwall, Thomas

Keywords

anomaly detection, outlier detection, time series, software-defined networking, microservices

Other note

Citation