Abstract
This paper surveys the literature relating to the application of machine learning to fault management in cellular networks from an operational perspective. We summarise the main issues as 5G networks evolve, and their implications for fault management. We describe the relevant machine learning techniques through to deep learning, and survey the progress which has been made in their application, based on the building blocks of a typical fault management system. We review recent work to develop the abilities of deep learning systems to explain and justify their recommendations to network operators. We discuss forthcoming changes in network architecture which are likely to impact fault management and offer a vision of how fault management systems can exploit deep learning in the future. We identify a series of research topics for further study in order to achieve this.
Highlights
The pressure to achieve greater data rates from limited radio spectrum resources is driving changes in cellular network architecture as 5G evolves
A decentralised Radio Access Network (RAN) architecture has emerged where groups of small, densely deployed cells are associated with a single macrocell, with signalling transmission retained by the macrocell but user traffic largely devolved to the small cells [1], [2] and [3]
SUMMARY we have described the principal machine learning (ML) techniques which have been used in cell fault management, and for each group of techniques we have explained in broad terms which activity within fault management the techniques are most applicable to
Summary
The pressure to achieve greater data rates from limited radio spectrum resources is driving changes in cellular network architecture as 5G evolves. This will require complex strategies to adjust cell coverage and turn base stations off and on again depending on traffic levels, without disruption to users [6] These additional capabilities included in the new RAN architecture will require it to have many more configurable parameters than previous generations [7], [8] and [9], whose settings may vary according to local conditions. Algorithmic approaches can be very effective but lack the transparency of rule based systems and are typically limited to a narrow problem area, requiring the use of diverse algorithms to cover the complete fault management problem space, each requiring input from both network domain experts and algorithm specialists to set it up and maintain it. We list areas for future research to close these gaps
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
