A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis

The detection of cells with sub-optimal performance and the identification of the root-cause of such performance is a crucial and challenging task in Network Performance Management (NPM). The contemporary NPM approaches, being reactive, silo-based, and highly expert-reliant, are not viable options f...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Ali Rizwan, Mona Jaber, Fethi Filali, Ali Imran, Adnan Abu-Dayya
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	ZSM zero-touch mobile network service management performance management CDR Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/b13cad6056fd419b9f74fdcef9896504
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:b13cad6056fd419b9f74fdcef9896504
record_format	dspace
spelling	oai:doaj.org-article:b13cad6056fd419b9f74fdcef98965042021-12-03T00:00:49ZA Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis2169-353610.1109/ACCESS.2021.3129281https://doaj.org/article/b13cad6056fd419b9f74fdcef98965042021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9620089/https://doaj.org/toc/2169-3536The detection of cells with sub-optimal performance and the identification of the root-cause of such performance is a crucial and challenging task in Network Performance Management (NPM). The contemporary NPM approaches, being reactive, silo-based, and highly expert-reliant, are not viable options for such tasks anymore, particularly in the emerging complex heterogeneous mobile networks. The state-of-the-art research in the field of data-driven Artificial Intelligence (AI) is a ray of hope for developing innovative solutions for such NPM tasks. However, the scarcity of holistic and detailed real network data limits the potential of this approach. In this study, we present a comprehensive AI-driven framework for the auto-diagnosis of cells with sub-optimal performance in a real network. We have explored and shared insight about an untapped comprehensive Call Detail Record (CDR) dataset from a real network operator. The outcome is anonymous and annotated data made public to encourage further research in this domain. We employ a <inline-formula> <tex-math notation="LaTeX">$K-means$ </tex-math></inline-formula> clustering method that exploits CDR data and domain experts’ input for the identification of particular types of cell performances. Next, a support vector machine-based classifier is developed for real-time applications which classifies the network nodes based on their performance with an accuracy of 97.69%. Subsequently, we introduce an algorithm that uses the classification results for the root-cause analysis of sub-optimal performance by leveraging network topography and area knowledge. The method succeeds in reaching the outcomes of an expert-led root-cause analysis and beyond. At the same time, the algorithmic approach limits the manual root-cause analysis to 30 possible scenarios per hour as opposed to analysis of 759 cells, thus it reduces the workload of an expert significantly. In the broad picture, the proposed AI framework lays the foundation towards zero-touch mobile network and service management starting with automated NPM and root-cause analysis.Ali RizwanMona JaberFethi FilaliAli ImranAdnan Abu-DayyaIEEEarticleZSMzero-touchmobile networkservice managementperformance managementCDRElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 157699-157714 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	ZSM zero-touch mobile network service management performance management CDR Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	ZSM zero-touch mobile network service management performance management CDR Electrical engineering. Electronics. Nuclear engineering TK1-9971 Ali Rizwan Mona Jaber Fethi Filali Ali Imran Adnan Abu-Dayya A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis
description	The detection of cells with sub-optimal performance and the identification of the root-cause of such performance is a crucial and challenging task in Network Performance Management (NPM). The contemporary NPM approaches, being reactive, silo-based, and highly expert-reliant, are not viable options for such tasks anymore, particularly in the emerging complex heterogeneous mobile networks. The state-of-the-art research in the field of data-driven Artificial Intelligence (AI) is a ray of hope for developing innovative solutions for such NPM tasks. However, the scarcity of holistic and detailed real network data limits the potential of this approach. In this study, we present a comprehensive AI-driven framework for the auto-diagnosis of cells with sub-optimal performance in a real network. We have explored and shared insight about an untapped comprehensive Call Detail Record (CDR) dataset from a real network operator. The outcome is anonymous and annotated data made public to encourage further research in this domain. We employ a <inline-formula> <tex-math notation="LaTeX">$K-means$ </tex-math></inline-formula> clustering method that exploits CDR data and domain experts’ input for the identification of particular types of cell performances. Next, a support vector machine-based classifier is developed for real-time applications which classifies the network nodes based on their performance with an accuracy of 97.69%. Subsequently, we introduce an algorithm that uses the classification results for the root-cause analysis of sub-optimal performance by leveraging network topography and area knowledge. The method succeeds in reaching the outcomes of an expert-led root-cause analysis and beyond. At the same time, the algorithmic approach limits the manual root-cause analysis to 30 possible scenarios per hour as opposed to analysis of 759 cells, thus it reduces the workload of an expert significantly. In the broad picture, the proposed AI framework lays the foundation towards zero-touch mobile network and service management starting with automated NPM and root-cause analysis.
format	article
author	Ali Rizwan Mona Jaber Fethi Filali Ali Imran Adnan Abu-Dayya
author_facet	Ali Rizwan Mona Jaber Fethi Filali Ali Imran Adnan Abu-Dayya
author_sort	Ali Rizwan
title	A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis
title_short	A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis
title_full	A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis
title_fullStr	A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis
title_full_unstemmed	A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis
title_sort	zero-touch network service management approach using ai-enabled cdr analysis
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/b13cad6056fd419b9f74fdcef9896504
work_keys_str_mv	AT alirizwan azerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT monajaber azerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT fethifilali azerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT aliimran azerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT adnanabudayya azerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT alirizwan zerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT monajaber zerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT fethifilali zerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT aliimran zerotouchnetworkservicemanagementapproachusingaienabledcdranalysis AT adnanabudayya zerotouchnetworkservicemanagementapproachusingaienabledcdranalysis
_version_	1718374008055922688

A Zero-Touch Network Service Management Approach Using AI-Enabled CDR Analysis

Ejemplares similares