One of the driving assumptions behind 5G is that it will be deployed hand-in-hand with the technologies that will result in a highly automated network, such as AI-enabled network analytics and virtualization. The resulting cost savings, and improvements in efficiency and quality of service, should help to justify the heavy expenditure in deploying yet another generation of mobile networks.
This justification is much needed, at a time when 4G still has plenty of life in it for most operators and most locations, if all they needed was to add capacity and support ever-higher video resolutions. But they need more than that, notably an entirely new cost base for supporting existing services, and a flexible platform capable of enabling a huge diversity of new applications. That diversity will require millions of connections to be optimized on the fly, which will only be done with high automation.
None of this is new. Operators have talked for a decade about the benefits of greater automation in their networks. While manual processes have been increasingly removed from many customer-facing operations, progress has been slower in the network itself, especially in the areas examined in this report – planning, deployment, optimisation and maintenance of the RAN and the end-to-end network (including core and transport).
This has been for three broad reasons:
- Immaturity in the technology, which has led to fears that operators would end up automating bad practice, or sufferings trade-offs in the quality of their networks.
- Concerns that the organisational upheaval, including loss of staff, would outweigh the benefits.
- Lack of urgency. While most operators assumed, even in the early days of 4G, that there were cost and efficiency benefits to be derived from automation of the network, these were not seen to be sufficiently dramatic to justify significant investment and disruption.
This is changing as operators start to embark on 4G expansion and 5G roll-out. Three main facets of those processes are changing the economics of the network and making RAN automation more critical to the mobile business case. Those three are:
- Densification with small cells, from indoor picocells to outdoor mini-macro base stations.
- More complex antenna arrays, including Massive MIMO systems with advanced beamforming to enable 5G to operate well in high frequency spectrum.
- Virtualization of the network, beginning with the packet core and gateways, in most cases, but as 5G evolves, extending to the RAN.
These trends introduce a huge number of new components – from thousands of physical small cell sites to hundreds of dynamically provisioned virtual network functions (VNFs), to millions of connected devices – into the network. This means modern networks are becoming hard for humans to understand, predict or manage.
Increasing levels of virtualization of many elements of the network, including the RAN means all the VNFs will be managed and coordinated with end-to-end orchestrators and SDN (software-defined networking) controllers. This end-to-end virtualized network will lend itself to high levels of automation, which will transform the total cost of ownership (TCO) of the RAN, core and backhaul.
Therefore, the fully automated, ‘self-driving telco network’ is a one of the cornerstones of the next generation vision for many operators, since it would keep track of all these moving parts, enabling them to be allocated and provisioned optimally and efficiently for each user and application. That would also dramatically improve operating costs and quality of service.
As for the three main factors which have slowed uptake so far, some solutions are emerging. Technologies are becoming more intelligent, to address the fears that automation can be dumb – more advanced self-optimizing network (SON) capabilities and the application of machine learning (ML) to network planning and maintenance are examples.
And the fears of organizational upheaval are being offset by rising urgency – automation will not be a luxury in 5G, but a necessity, and its predicted impact on the operating costs of the network are expected to be significant.
A consensus prediction among the 83 MNOs surveyed for this report was that operating expenditure on the network (opex) needed to fall by an average of 33% over the first five years of 5G deployment, and that about 30% of those savings would come from RAN and transport automation. That equates to an 11% fall in opex. By contrast, in the first five years of 4G, they believe RAN and transport automation would have had the potential to reduce total opex by only about 3%.
The potential benefits are well understood, but there are still many challenges. Indeed, compared to Rethink’s previous survey related to automation, conducted in the first quarter of 2018, there is a higher degree of short term caution about implementing automation than there was then. Operators’ timelines for significant degrees of automation have lengthened, though their goals for 2025 are similar or even more ambitious.
This partly reflects a natural dose of realism that creeps in when the deadline draws near. Automation will be a gradual process for most operators and that is often the best approach, provided each stage delivers benefits and is related both to short term economic goals and to future strategy.
However, the onset of caution also raises the risk that operators will move too slowly, and fail to plan effectively even if they do not need to implement the new processes and technologies immediately. The ultimate goal, according to early movers like AT&T and Deutsche Telekom, is to automate at least 80% of physical network processes and virtual network functions in the RAN and the end-to-end network. That will be a long and disruptive process, which needs to start quickly, or operators risk sacrificing many of the promised TCO and efficiency gains of their 5G roll-outs.
This article is an extract from Rethink Technology Research’s latest RAN Service report, entitled ‘RAN automation is central to the 5G case – but is it a distant dream? Automation and SON deployments and trends 2018-2025’. For more information, please contact John Constant on [email protected].