Network slicing is a crucial concept in the 5G landscape; indeed, for many operators it will be the best way to justify 5G investment at all. Yet it has remained a vision of vendor labs and international R&D projects, and very little attempt has been made to assess its economic potential. So while BT and Ericsson have a sharp axe to grind in this area, they have at least offered some financial analysis in their new report on the subject, claiming slicing could boost operator revenues by over one-third compared to non-sliced environments.
The concept is that a service or individual will be allocated a virtualized slice of network capacity, fully optimized for its particular needs. In a virtualized, software-driven network (SDN), an orchestrator will call up the combination of virtual network functions (VNFs) that a service would require, together with the appropriate piece of radio connectivity and spectrum, supporting capabilities such as high capacity, low latency, security and so on.
The modelling exercise by Ericsson and BT took place over four months, and compared network slicing with two alternative approaches for new service deployments – which they label in the report as “one big network” and “separate specialized networks”. (The study was limited to evaluating the impacts of slicing in the core network, and BT admits it has not yet addressed the challenging but essential element of slicing the air interface)
“We found that over a five-year period, introducing new services by using network slicing and operational automation generated 35% more revenue than by using one multi-service network,” Marielle Lindgren, head of Ericsson UK, said in an interview. “The revenue increased 15% when compared to several networks with dedicated resources, demonstrating how the technology enables market stimulation, faster time to market, and opportunities from smaller niche services.”
The base case involved ramping up to 40 annual service launches during a five-year period, all of them focused on IoT – the study was assuming that, as many MNOs are indicating, they will continue to enhance their networks for their core business model of mobile broadband, but incremental revenues will come from the IoT and enterprise services. So it is here that there would be the greatest return on investment from a new architecture, since these would be entirely new revenue streams. Of the 40 new services BT and Ericsson envisage, 60%, in their model were for critical machine-type communications and 40% for massive machine-type communications – the two categories into which the companies, following general industry thinking, divide IoT services.
Network slicing also delivered about 40% reduction in opex, based on the assumption that operators would introduce automation for provisioning and managing the slices, which was not assumed in the other scenarios. This all added up to an overall impact of 150% increased economic benefit, said the report authors. It was clear that, with 40 services a year, they could not all be managed and orchestrated manually.
Jan Häglund, head of architecture and solutions at Ericsson, explained to Mobile Europe: “We modelled in an investment of automation so the effort to instantiate a new slice is minimal, and the modification of the characteristics of the slice, for example the scale or capacity, should not pass by the operator or service provider; you should be able to do it directly by the tenant, such as the car manufacturer.” That also, of course, enables operators to avoid accusations of violating net neutrality since the customer, not the MNO, decides on the characteristics of the slice.
Other advantages include shorter time to market for new services, more efficient use of resources, and more tailored services for high value customers like enterprises or utilities.
The report accepts that there are large overheads in deploying such a platform, but claims the operator will typically see return on investment within two years, based on providing 40 new automated services. Even with just five new services, ROI should be achieved about three years after commercial launch.
Häglund said that the huge numbers of devices envisaged for the IoT would create the need for a flexible network that could serve different use cases. He said: “They will range from a device sending data every month, say in massive machine type communications, a sensor network, agriculture, parking meters or whatever your imagination comes up with. Or it could be ranging all the way up to very critical machine type communication that maybe isn’t even possible in today’s networks – very low latency for consumers like in virtual reality or in gaming, or in the industrial segment like robotics control or automotive. All in all, the requirements are growing so it begs the question of what enabling technology you need to cater for this.”
Early slicing demonstrations were little more than VPNs, but recent ones have been far more sophisticated, although BT admits it has not yet tackled the thorny issue of slicing the air interface itself. This will be an essential challenge to address, to achieve a fully end-to-end slice, drawing on resources from transport to core to RAN to storage to processing power.
Operators are also divided about whether the slices will be static and few in number (one for connected cars, perhaps, one for mobile broadband, and one for high availability services such as emergency response).
Initially, they are defining slices as dedicated functional layers hosted on shared physical resources, but there might only be four or five such layers, and they would not change their key characteristics often.
The real vision, though, is for many slices, each one called up and orchestrated dynamically for a particular service or, eventually, even for an individual user. The latter is supported in 5G R&D projects like Telefonica’s and Huawei’s User Centric No Cell (UCNC) architecture, in which a cell of mobile capacity effectively forms itself around a subscriber on-demand and follows them around for as long as they need it.
That sort of approach is not as far away as some think, in technical terms at least. Artemis and its Personal Cell idea has been pursuing similar concepts even before 5G.
Such approaches will give the operator real flexibility to support huge numbers of enterprises, MVNOs or even consumers in an optimized way, but only allocating the resources they really need, rather than over-provisioning the network for all.
But there are several stages to go through before this fully programmable, dynamic and responsive network becomes a reality, and an important one is to test and prove the investment case and the commercial return. In that respect, the new report is a valuable contribution.
It also highlights the urgent need for MNOs to look beyond their core mobile broadband and consumer businesses, and to find a far easier and more cost-effective way to deliver a wide range of new revenue-generating services, before alternative wireless providers – harnessing open source networks and shared or unlicensed spectrum – steal the low hanging fruit.
So Häglund pointed to the great hope that rests on slicing, an efficient way for MNOs to secure significant enterprise business, looking beyond just connectivity, after many years of trying – and to extend that into the new use cases of the IoT.
“If you are struggling with growth in the consumer segment, which is very big but very challenged in terms of competitiveness, I think many of our customers could broaden their addressable market by looking to adjacent industries,” he said. He is clear that the barriers to diversifying the business model, for all but the largest MNOs, have prevented operators reaching beyond the consumer segment. “The current paradigm is one big network with one operation but different subscribers. You’re not really able to tailor the service to a new sort of industry. If you want to react to something new, then you will always have to ask whether it’s worth it to upgrade your network for this opportunity. In many case, you’ll come to the conclusion ‘nah, let’s wait and see’.”
To drive these operators to take the plunge and target new markets, economic justification will be important, but so will standards, which reduce cost, enable interoperability and cut risk.
“We’re positioned to bring millions of devices onto our networks as the IoT ecosystem grows,” said Maria Cuevas, head of mobile core networks research at BT. “This study gives us guidance as to where our investments will achieve the best results. The more services we deploy with network slicing, the greater economic benefit we will see, enabling us to better serve our customers. In order to achieve this vision, it is important that the industry provides cost-effective solutions to support end-to-end orchestration and adds automation to the operations and management of network slices.”
This summer, the 3GPP took a step forward in this direction. Its SA5 Group entered the normative phase of its work on the management and orchestration of network slicing, having completed an in-depth study. This group works on the system architecture which will develop alongside the RAN itself (the 5G radio also includes specific features to support slicing of the air interface and enable end-to-end slices across mobile and fixed links).
Also, the ITU (International Telecommunications Union) has put software-driven network orchestration, with a focus on slicing, at the top of its own agenda. It has set up a Joint Coordination Activity group to focus the work of standards bodies which hope to support the final IMT-2020 (ITU nomenclature for 5G) approved standards when these are chosen this year. It also hopes to include the efforts of the open source community.
The aim is to promote standards harmonization and prevent unnecessary duplication of effort, and so accelerate progress towards a commonly deployable platform. The JCA should provide a place where different players can contribute the work that is going on in standards bodies and open source groups across the ecosystem. It will focus on how slicing, and what ITU calls “softwarization”, will result in deeply programmable networks with the agility to support the specific requirements of any particular 5G application.
“Here we see the fundamental importance of software-driven network orchestration,” said Chaesub Lee, director of the ITU telecommunication standardization bureau. “5G systems will be required to support an enormous number of diverse ICT applications. Our aim is to ensure that every application is able to perform to its full potential.”
“Network management needs be able to take a look at the network, create slices of the network, ensure isolation, ensure security, and ensure that you can build these systems so that you can have orchestrated pieces that drive the controller pieces that then feed into the thing that most people really care about,” said Scott Mansfield, SG13 vice-chairman (from Ericsson). “That is how you make money out of the network.”
There are many 5G projects round the world which are focused on aspects of slicing. One of those under the auspices of the European Union’s 5G PPP initiative is Monarch, a successor to the recently completed NORMA project. Both of these were led by Nokia, but Huawei has just announced that it is also joining Monarch. The 5G-Monarch consortium also includes Deutsche Telekom, Samsung and Telecom Italia, as well as several leading research institutions.
The goal of the 5G-Monarch consortium is to enable 5G use cases by designing a flexible and adaptive architecture including slicing. Work items include a cloud-enabled protocol stack, inter-slice control and management, and experiment-driven optimization along with design of secure, resilient and elastic network functions.
“Network slicing is not only a technology enabler, but a means to revolutionize the telecommunications ecosystem and truly realize the vision of 5G to change society,” said Zhou Hong, president of the Huawei European Research Institute. “The flagship 5G-Monarch project has thus placed network slicing at the core of its flexible and adaptive architecture design. We remain fully committed to 5G PPP and look forward to shaping the wireless future together.” The project will run until 2019 and the European Union is providing €7.7m in funding.
Once defined, the architecture will be deployed in two testbeds based on scenarios with demanding network requirements. The first will be at a sea port with a requirement for secure, reliable communications, and will be hosted by consortium partner the Hamburg Port Authority in Germany. For each testbed, 5G-Monarch will instantiate the architecture and extend it with use case specific functionality, such as resilience and security in this case.
The second test deployment will be in a busy tourist city center, as yet unnamed but expected to be in Germany or Italy because of the MNOs involved. The specific functionality in this case will be resource elasticity, to make an efficient use of network resources when mobile broadband demand is high.
The consortium partners say the project will feed its results into commercial developments by members and others, and expects it to affect the evolution of orchestrators and edge-cloud RAN, as well as new vertical market services enabled by slicing.
Huawei was also extolling the benefits of slicing at last week’s New Generation Internet Infrastructure Forum in Beijing, where it talked about a project focused on 5G power slicing, working with China Telecom’s Beijing Research Institute and China Electric Power Research Institute. These are investigating how slicing technologies can benefit the power industry with applications such as automatic power distribution. The three companies aim to prove that a 5G slice can achieve security and isolation at the same level as those provided by a private power grid, but with lower cost and higher flexibility.
“The three-party collaboration project is the first exploration of 5G slicing in power industry applications. 5G slicing is applied to vertical industries, such as the electric power industry, and this will incubate more new applications and business models,” said Qiu Xuefeng, VP of packet core networks for Huawei’s Cloud Core product line. “The core network is critical to implement 5G slicing features, such as on-demand network definition, fast deployment, automatic operation, end-to-end SLA assurance, and capability exposure.”
BT demonstrates a drone network slice:
Last month, BT demonstrated a drone, operated by a dedicated slice within its network, at its 5G slicing testbed in its labs in Suffolk, east England.
The operator worked with Ericsson (though it also has slicing trials with Huawei) and with drone company uVue.
Maria Cuevas, BT’s head of mobile core networks research, said: “The demonstration was most effective in showing that we are able to protect a private ‘slice’ of the network for uVue’s drone as the rest of the network was deliberately overloaded. The slice allocated to drone control remained perfectly functional and stable, enabling the drone to remain responsive to the pilot’s control at all times.”
Over 100 flights were executed, aiming to prove that this is a viable method of drone control, with the advantage of allowing the unmanned vehicles to operate with existing air traffic control systems. For this demo, HD video was streamed from the drones across the pre-5G test network at low latency, and the video was then displayed to an audience.
Russ Delaney, director of tech ops at uVue, said in a statement: “I was delighted at how responsive the drone control was over the mobile network and at how reliable the private slice was. Delivery of real time low latency HD video footage back over the network provided a completely new standard for ‘eye in the sky’. This is a key milestone in UAV development, showing that cellular networks have immense potential to provide uncompromised drone control and hence air safety.”