With revenues from consumers expected to be flat over the next few years, operators are forced to focus increasingly on enterprises to recoup investments in 5G spectrum and infrastructure. Network slicing is seen as critical in generating these revenues by enabling differentiated services that appeal not just to enterprises, but also some consumers in gaming and premium entertainment for example.
The primary vendors have all promoted network slicing and invested heavily in tools to help implement it and ensure that individual slices can be monetized effectively through integration with existing billing and CRM (customer relationship management) systems.
The ability to configure slices for individual enterprise customers or use cases quickly and efficiently is also critical, and Nokia was quick off the blocks here with its announcement in October 2020 of automation capabilities incorporating network management, controller and orchestration capabilities. Nokia claimed this enabled mobile operators to deliver and assure network slicing services within minutes instead of hours or days, although such capabilities are also now available from its competitors.
Nokia did follow up in January 2021 with announcement that it had piloted network slicing with Saudi Arabian telco Mobily for delivery of fixed wireless access (FWA) to both consumer and enterprise customers. In this case, slicing enabled the infrastructure to be split by application with voice, data, online gaming and home office applications each allocated their own end-to-end resources and guaranteed QoS as appropriate for each one. Nokia reckoned this demonstrated the capabilities of slicing for provision of differentiated public services and the need for end-to-end orchestration spanning RAN, core and overall data transport.
That package was also significant for combining LTE, 5G Non-Standalone (NSA) and 5G Standalone (SA) operation, underlining that slicing is not solely for 5G. It can provide the benefits to 4G and be introduced to 5G through NSA deployments.
More recently, Ericsson has been raising consciousness after amassing a comprehensive portfolio of slicing tools and capabilities, issuing a slew of papers and blogs. The vendor has two key messages – firstly that slicing offers prospects of major revenue gains for operators; and secondly, it is not a straight binary choice for enterprises, between using slices within public networks and establishing their own private networks to meet new demanding use cases. Both approaches can coexist with private networks most appropriate in some situations, perhaps to suit dedicated isolated applications, and slicing in others.
Indeed, Ericsson repeatedly hammers home the argument that slices within public networks are more suitable when wide area coverage is also required. Although private networks can be integrated into public cellular services for this to happen, this rather defeats the point of them. Yet if they are standalone their benefits are confined locally, while if they are integrated the same capabilities can be provided more readily and just as securely through network slicing.
Ericsson also contends that the time and resources required to deploy a standalone private network are beyond the reach of smaller enterprises and there are already signs this will be the case. Most private networks deployed so far have been by major automotive and manufacturing multinationals such as Bosch and Volkswagen.
Ericsson conducted separate research to back up both its assertions about network slicing, firstly the potential revenue generation and secondly the appeal to enterprises. For the first, it conducted a study with Arthur D Little (ADL) entitled ‘Network slicing: A go-to-market guide to capture high revenue potential’, which identified $200bn in potential value for service providers by 2030. This would be about 30% of the total $510bn annual revenue generated by mobile services by then, representing 36% CAGR once the field gets going.
By 2025 the total annual revenue from network slicing for operators would be $45bn on this basis, which sounds a rather optimistic figure, while 2030 is too far away for any confident prediction, with large margin for error.
Nonetheless, ADL was confident that take-up of network slicing will be substantial across most major vertical segments. It forecast that the top six industries will generate 90% of slicing revenue for operators. Although manufacturing is often touted as the biggest opportunity for private networks, ADL puts down healthcare as the largest for network slicing as a whole, accounting for 21%, followed by government on 17%, transportation 15%, energy 14%, with manufacturing only fifth on 12%, and media/entertainment on 11%. Of the rest, just three register – financial services on 6%, retail surprisingly low on 3%, and agriculture climbing from a low base to 2%.
Not all of these network slicing deployments will feature advanced use cases calling for ultra-low latency, for example. Some will just support general mobile services for given enterprises, which partly explains why manufacturing does not rank more highly.
Indeed, another study – where Ericsson collaborated with Analysys Mason – found that around 80% of enterprises plan to use a 5G private network of some sort, but also that this did not preclude network slicing as well. In fact, 50% of enterprises plan to deploy both private 5G network slices managed by operators and private 5G networks. That too gives a clear indication of potential for network slicing.
Huawei is naturally also keen to harvest network slicing but has avoided some of the headier rhetoric. The company recognizes that there is a lot of heavy lifting to be done before early trials and tentative deployments lead to widespread commercial availability.
This reflects a wider view in China that network slicing is still at the stage of proprietary implementations and cannot be said to have come of age until it reaches full vendor interoperability. This view was espoused recently by Zhang Jing, standards engineer at China Information and Communications Technology Institute (CAICT), a government-run thinktank. “There is still a certain gap between network slicing and commercial deployment,” Zhang said at a recent conference. “The industry has a long way to go, and requires the efforts of all players in the industry chain.”
Lacking still, according to Zhang, are clear standards for interoperability in key slicing processes, such as coordination between access, transport and core networks, or in automating end-to-end deployment. These are the sort of capabilities Nokia, for one, has been trumpeting, but that is within a single vendor walled garden.
This will have to be resolved if network slicing is to fit with Open RAN multivendor deployments. On the surface, slicing is well catered for in a virtualized Open RAN world. It is, after all, enabled by virtualization, with the separation between high level control plane and underlying network resources. In a virtualized Open RAN network, it should be possible to add network functions on demand and relinquish them subsequently when not required any longer.
As for RAN resources, the Open RAN Alliance claims to have prepared for differentiated services under network slicing by defining the RAN Intelligent Controller (RIC), which is split between a near-real time component to operate with latency between 10ms and 500ms, and a non-real time RIC for non-time critical tasks.
However, the radio resources at the edge of the physical network cannot be conjured into existence at a whim and are limited by the remote radio heads and antennas that have been deployed. There will always be some lag then, unless there is over-provisioning within the RAN, which incurs extra costs. There will always be some trade-off between cost and guaranteed QoS for all services, which in practice means that users will go on experiencing variations in QoS for less critical tasks.
Huawei’s chief architect of network transformation, Andy Li, has argued that operators need to be thinking now about network slicing, with a view to making it fully competitive only after a few years, by around 2025. That happens to be the year Ericsson is looking at for slicing to get fully into its stride, so perhaps there is some convergence of views. The main point is that operators are being urged to conduct trials and evaluations now.