The enhancement of network operations centres operating models through the use of data analytics

Abstract

Telecommunications is an essential service as it enables people to communicate and supports business to function efficiently. This service demands that high availability is upheld continuously. Some network faults can be avoided by implementing preventative maintenance; however, certain faults cannot be avoided and must be handled efficiently when they occur. A Network Operations Centre (NOC) executes the role of detecting and analysing faults to identify the root cause. Network Operations Centre engineers perform analyses manually, which adds to the time duration of the fault. An increase in fault alarm volume is the underlying cause of long fault analyses. This Design Science Research study aimed to examine data analytics solutions that enable network engineers to perform optimally during network outages. The three focus areas of the study were i) manual network analysis, ii) incident management analysis, and iii) the analysis of a smart grouping mechanism for alarm patterns. The collaborative rapid application development approach was adopted to introduce agility into the design, development and demonstration processes. Each of the three focus areas of the problem were handled as an independent delivery. Artefact development was based on data collected from users, alarm data, and incident data. The main finding of the study is that the rule set based algorithms used in NOC automation are ineffective and cause long-term damage by hiding useful data needed for decision-making. A data analytics artefact developed in this study introduced 97% efficiency in alarm analysis and enabled unsupervised machine learning (ML) capability, allowing users to gain insight into the network with minimal effort. The proposed NOC advanced Information System is the operational model recommended for Mobile Network Operators.