There may already be more commercial 5G networks in the world than anyone had expected by mid-2019, but none of these, not even the wildly popular SK Telecom network in Korea, stands up to our definition of ‘true 5G’. All early adopters have led off with the 5G New Radio Non-standalone (NR NSA) platform, which means they can accelerate progress by deploying a 5G RAN, but retaining the LTE core. In most cases – with some exceptions in the South Korean operators’ heavily customized networks – the operators are rolling out conventional base stations, not Cloud-RAN, and even if their LTE cores are partially virtualized, they are certainly not cloud-native.
To claim a ‘true 5G’ network – rather than a souped-up 4G one using higher capacity spectrum – operators will need to tick several boxes. These include the next release of 3GPP standards, Release 16, which will introduce enhancements to latency, availability and performance; an NR Standalone roll-out with a cloud-native 5G core; and an open applications platform which can deliver significantly different user experiences to those of 4G, not just more of the same at higher speed.
Those developments will enable operators to support flexible allocation of spectrum and network resources according to the requirements of a particular user or service, increasing their own cost efficiency and the quality of experience for customers. And they will lay the foundations for full end-to-end network slicing, in which virtual networks will be automatically created on-demand, collecting fixed, mobile and cloud resources from multiple sources, to the precise requirements of a particular application, and charging on a marketplace basis.
Not all operators will go this far, and for those which do plan fully dynamic slicing, it will be a journey of many years. But the foundations to build a 5G network that can support limitless numbers of different services, each with different requirements, can start to be laid now, and a few operators are showing the way. Some are established names which are reinventing themselves to re-ignite growth, like AT&T. Some are disruptive new entrants like Reliance Jio and Rakuten, with the luxury of a greenfield deployment. Some are challenger MNOs, like Three UK, which aim to harness the new technologies to improve their agility and their differentiation from their larger rivals, in order to steal market share.
Established MNOs will need to be embarking on strategies to reduce cost, increase commercial agility and support new industries and use cases, otherwise others will do it for them. The rising interest in private spectrum in countries like the UK and Germany (see separate item below), and the upcoming definition of 5G standards for shared spectrum, mean that MNOs will not be the only deployers of the new networks. They may find themselves competing with their own vendors, Nokia in particular, to offer cloud-based enablement for specialist service providers and industrial MVNOs.
And as cloud, and particularly edge, computing becomes more integrated with connectivity, they will need to fend off the webscale giants too – or work out a way to partner with them profitably, before AWS, Microsoft, Facebook and Google get any more interested in controlling networks themselves (in the latest development in that area, Facebook is conducting trials in the USA’s CBRS shared airwaves, following the example of Google and Amazon).
Akraino releases first edge blueprints, opens up RAN control to third parties
The combination of mobile connectivity and edge computing could support or enhance many of the applications that are commonly assumed to be ‘5G’, from virtual reality gaming to smart factories.
But there are difficult decisions for telcos when they decide on edge strategies. One is whether to invest in edge infrastructure themselves, especially for services that require nodes in locations that do not align with telco sites such as central offices or towers. Alternatively they can work with third parties, but risk having their role in the value chain squeezed to that of a dumb pipe.
Several large telcos, including AT&T and Deutsche Telekom, are looking to harness a combination of open source systems and 5G to establish platforms in which operators can take a leading role, even in a broad ecosystem. DT has its Silicon Valley-based MobiledgeX spin-off, while AT&T has placed its Akraino Edge Stack technology into open source, and is driving integration with the other open platforms it has spearheaded, such as ORAN (Open RAN).
Akraino Edge Stack is now part of the Linux Foundation’s LF Edge program, an umbrella for several edge-related projects including EdgeFoundry. It aims to produce a full open source software stack that supports a high availability cloud architecture optimized for edge computing. It will target three types of edge platform – enterprise edge, OTT edge, and telecoms edge.
LF Edge recently released Version 1.0 of the Akraino Edge Stack, which is based around 11 families of blueprints for different use cases and scenarios. These include telco appliances for the RAN; telco appliances for the network cloud; application deployment on ARM servers; enterprise edge use cases at high-density locations; Kubernetes-native infrastructure; and an IoT gateway.
The first key element of the Telco Appliance for RAN blueprint to be unveiled has been co-developed by AT&T and Nokia and will be placed into open source via Akraino. This is a radio edge cloud (REC) appliance which creates an open software environment specifically for the control and management of distributed RANs. By opening certain elements of the RAN up to third party control, it could make distributed mobile networks more open and multivendor, and in turn, improve connectivity to the edge nodes, for instance to support specialized requirements such as real time response.
The REC would remove the control of the network from proprietary systems embedded at the heart of vendors’ locked-in platforms, and open it up to third parties via open APIs (application programming interfaces). This could have a radical impact on the cost of mobile networks, and the operator’s ability to make its own decisions about how to manage its RAN, and the best tools for the job.
According to Michael Murphy, CTO North America for Nokia, the initial use case for the REC will be to support a near-real time RAN Intelligent Controller (RIC). This highlights the interplay between the various Linux Foundation projects based originally on AT&T code – the RIC concept was developed by the ORAN Alliance as part of their work on a set of common interfaces that would allow equipment from multiple suppliers to interoperate within open, virtualized mobile networks, so improving price and innovation competition among vendors.
The RIC network element will control some aspects of the RAN and will be an open platform, taking network controls away from the vendors’ proprietary, locked-in systems. The REC would open up the RIC to third party apps, (called xApps in the ORAN world), which would provide management and control functions for the RAN. It will support automated configuration and integration testing of the full stack starting from below the operating system layer and moving up to the RIC. The REC will share many hardware and software components, including installation, configuration management and APIs, with other family members.
The aim is be to enable operators to select companies with particular expertise in one aspect of RAN management, such as as video load balancing, and mix and match their functions via an open app platform.
It may seem surprising that Nokia is endorsing a system which will help to break down its ability to lock operators into its architecture, and reduce the mystique vendors have placed around RAN control. Indeed, Murphy was not wholeheartedly in support, saying that opening up the radio’s edge in this way would not work all the time. “We thought there was merit in this and there was value in accelerating innovation and reducing costs,” he said. “But there are some use cases that we don’t agree with.”
Over time, the REC could be extended to other use cases beyond the RIC. These might be other aspects of the open RAN, or to enable third parties to develop applications which rely on mobile connectivity, providing them with access to those RAN resources through open interfaces. This would maintain the early association of edge with mobile connectivity – something that was pushed heavily by ETSI MEC, but has been somewhat lost amidst the excitement of non-telco players, like the cloud giants, about edge computing.
Murphy said that it is not easy to predict all the use cases for the REC but that having an open source edge cloud with open interfaces to the RAN control will open up many new possibilities for operators and their suppliers and development partners.
Kandan Kathirvel, a director at AT&T and chair of the technical steering committee for Akraino, said: “The radio edge cloud is primarily focused on the RAN deployment. The 5G deployment is really going to be based on these RAN deployments because that’s where virtualization and the RAN intelligent controller is needed to really scale 5G.”
“From a telco perspective we believe that there are three things that they’re extremely well equipped with, and that’s what differentiates the telcos in this world of edge,” said Arpit Joshipura, general manager of edge, IoT, and networking at Linux Foundation. This trio comprises location, latency, and mobility.
The Foundation plans to update Akraino on a recurring six-month cycle with the next version including another five blueprints, Joshipura said. “I think our next phase, where each application is being looked at for a particular vertical, that’s when we start gaining a lot more traction and deployment — not just participation,” he added.
Plans are underway for Akraino Release 2, which will include new blueprints and enhancements to existing ones, tools for automated validations, defined edge APIs and new community lab hardware.
Contributing members to Akraino include ARM, AT&T, Ericsson, Huawei, Juniper, MobiledgeX, Nokia, NTT, Red Hat, Seagate and Wind River.
The first Akraino blueprints:
As well as the vRAN, the Akraino stack also aims to enhance the flexibility of edge cloud services in the 5G core, universal CPE (uCPE), SD-WAN and carrier edge media processing.
With Release 1.0, the Linux Foundation outlined six blueprints which are already “ready and proven”, including the REC.
Integrated Edge Cloud (IEC) Type 1 (Small Edge) and 2 (Medium Edge). These target telco applications with small and medium deployments of edge cloud based on ARM processors.
The Network Cloud family of blueprints for hardware configuration and fully automated deployment of multiple edge sites from a remote regional controller, including in 5G.
The StarlingX Far Edge Distributed Cloud blueprint for edge and ‘far edge’ use cases in high-density locations such as malls, airports and stadiums.
Edge Lightweight and IoT (ELIOT) supports use cases for IoT gateway and uCPE (SD-WAN), including Industrial IoT and smart cities where there is limited hardware capacity.
The Kubernetes-Native Infrastructure (KNI) Provider Access Edge blueprint leverages best practices from Kubernetes to manage edge computing stacks at scale on bare metal or public cloud.
“The Akraino Edge Stack release represents a significant milestone for open source collaborations in edge computing to reflect alignment across industry segments on common integration, validation criteria and tools for solution stacks,” said Imad Sousou, general manager of system software at Intel. “Blueprints, also known as common solution stacks, for network and edge cloud use cases will be essential for Akraino adoption at scale.”
Three UK steals a march with 144 MHz of 5G spectrum and cloud-native core
True to its name, Three UK became the third operator in the country to announce its launch date for commercial 5G, in August, and plans to leverage its unaccustomed spectrum advantage and its next generation core to steal a march on its larger rivals.
The Hutchison-owned company entered the market when it bought 3G spectrum in 2000, but despite a strategy to compete on network quality, in fact it largely grew on the back of very competitive pricing, enabled by its more efficient, 3G-only network. In 4G, that cost advantage diminished and Three slipped behind EE and Vodafone, in particular, in terms of its spectrum position; as well as being blocked by the regulator in its bid to acquire Telefónica O2 and therefore boost its scale and ability to invest in expansion.
However, it has been creative in its service propositions and its marketing, with something of the disruptive approach of T-Mobile USA’s Uncarrier policies. It has led the drive to unlimited data in the UK and claims its customers use 3.5 times more data than those on other networks.
And in 5G, it is in a far stronger position. Thanks to its £250m acquisition, in 2017, of UK Broadband, which owned significant assets in the 3.5 GHz band for fixed wireless use, Three is the only UK operator to have at least 100 MHz of 5G spectrum – the amount deemed necessary for some of the very high bandwidth applications that will set ‘true 5G’ apart. In fact it has 144 MHz across the 3.4-3.6 GHz, 3.9 GHz, 28 GHz and 40 GHz bands. It topped up UK Broadband’s midband assets with 20 MHz in the 3.4 GHz auction (though rivals got more), and the 700 MHz auction is still to come.
Most of these ‘true 5G applications’, outside vehicles, are not inherently mobile. Advances in virtual reality, for gaming and other processes, tend to be targeted to users who are sitting still with their smartphones or tablets. And one of Three’s initial focuses is on 5G fixed wireless access (FWA), allowing it to provide fixed/mobile bundles without a wireline network.
We believe FWA to be a niche case in the UK, where fast wireline connectivity is plentiful and cheap, despite limited fiber-to-the-home deployment so far (though that will change quickly). 5G fixed wireless may be useful to BT, to support the last few feet and lower the cost of getting fiber into the home. But for mobile-only players, it means being confined to ultra-rural areas or to low-budget groups like students. However, Three does have enough capacity to support FWA without leaving mobile 5G short-changed, and so it can look beyond the FWA niche, to a scenario where its mainstream 5G connections are so fast and reliable that users turn off their broadband line and use 5G for everything (without even needing a CPE).
In the first instance, Three will only offer FWA, and will limit itself to the home market in London. Several operators have taken this approach, most famously Verizon – going fixed first enables them to test the waters with the new technology without the challenges of supporting full mobility and sourcing smartphones. It also means the operator can deploy at a cautious pace at first since there are no coverage requirements.
By the end of the year, Three will also be offering mobile services, and will have reached 25 markets including Cardiff, Glasgow, Birmingham, Manchester and Liverpool. BT’s EE was the first to market, launching its 5G offering last week in six cities, and Vodafone will follow in seven cities on July 3. O2 will follow later in the year.
Three commented: “Our market-leading spectrum portfolio alongside investment in high capacity smart antennas will deliver the fastest 5G network in the UK. At launch, peak mobile speeds will be at least two times faster than other mobile network operators and will provide a more reliable connection and experience for customers.”
The 5G launches are part of Three’s £2bn ($2.5bn) 5G infrastructure investment plan, which calls for it to deploy 5G across 80% of its footprint by the end of 2022. It also says it will expand and extend its 4G network to boost LTE performance fourfold. Like most MNOs, Three sees 4G remaining the coverage network for many years, with 5G targeting particular hotspots of heavy usage or device density, or new industries.
Even more important than its spectrum advantage – it claims twice the amount of 5G spectrum of any other MNO thanks to UK Broadband and its recent auction purchases – is its cloud-native core. This is being implemented with Nokia and will help Three to be far more flexible and agile in how it develops new services and targets them at particular customer groups. By contrast, BT has an advanced cloud program, but has given no dates for its shift to a cloud-native 5G core (see Wireless Watch June 3 2019).
Three announced back in 2017 that it was working with Nokia to deploy a cloud-native core, and has become one of the world’s first MNOs to achieve such a platform for 5G. The 5G-ready system has been rolled out in 20 brand new data centers, connected by dark fiber, and has an initial capacity of 1.2TBps, three times bigger than its predecessor.
The operator says it is targeting security, scalability and cost benefits, above all, from its new core, which it has been using to support a trial network for staff. It hopes to extend the tests to selected consumers later this year, with a view to commercial deployment in 2020.
“In order to use the core network, we have to ensure that all of our mobile sites are connected to our new core. We achieved this milestone in December 2018. This means all of our customers will be able to enjoy the benefits of the new core network when it goes live,” said a spokesperson.
ORAN Alliance claims breakthrough in ending RAN vendor lock-in
The ORAN (Open RAN) Alliance aims to create a new stack reference design that will end vendor lock-ins and allow truly multivendor networks in the 5G era, with all the impact that implies on competition, pricing and innovation. It is claiming a major breakthrough on the way to that nirvana – a multivendor implementation for the X2 interface which connects two LTE base stations.
This is one of a set of essential interfaces between different network elements, which the ORAN Alliance is looking to make fully open and interoperable. In the past, RANs have remained semi-proprietary and locked-in, despite being based on 3GPP standards, because many key interfaces were implemented slightly differently by each vendor, making different suppliers’ boxes incompatible with one another.
The Alliance’s work has seemingly become more urgent because of the uncertainty over whether Huawei equipment will be allowed in 5G networks in some countries. Operators which have sourced their 4G RANs from Huawei are faced with a more difficult migration if they have to select a different supplier for 5G, because of the semi-proprietary nature of key interfaces such as X2, or the CPRI fronthaul protocol between radio heads and a centralized baseband.
Nokia and Ericsson have hastily come up with workarounds for operators which feel the need to migrate away from Huawei. Nokia offers an overlay which does not require X2 interworking between two vendors’ base stations, but instead introduces a new LTE radio from Nokia in a low band, purely to provide anchor network control functions as required in a 5G Non-Standalone architecture, which still uses a 4G anchor and core. The legacy 4G network still carries the LTE traffic, while the 5G network is deployed in a higher band such as 3.5 GHz.
Ericsson has something similar, and while these may be useful stopgaps for some operators, avoiding X2 altogether dodges many of the real issues. In the medium term, MNOs will be far more interested in an fully open and uniform implementation of X2, something which the ORAN Alliance claims is on its way. Arnaud Vamparys, SVP of radio networks and 5G champion at Orange, told TelecomTV that the Alliance has successfully completed an implementation test of the X2 interface linking base stations from Nokia and Ericsson.
That means an MNO would not be locked into its existing supplier when it moved to 5G, and it could add more 4G, or new 5G capacity with base stations from any vendor.
The Alliance has six technical workgroups focused on specific areas of the RAN and the essential interfaces, and has also released initial specifications for open fronthaul, based on work originally done by the xRAN Forum. This was an AT&T-initiated group, which merged with the Cloud-RAN Alliance to form ORAN.
Similar work is being done by the Telecom Infra Project’s Open RAN group, leading to speculation that these two initiatives will eventually converge. Both were very prominent at this year’s Mobile World Congress, where ORAN Alliance members showcased six proof of concept (PoC) platforms and the Open Fronthaul Specifications, which comprise control, user, synchronization and management plane protocols.
A long line of operators pledged support for Open Fronthaul, some saying they would introduce the specs to commercial networks from late 2019, and were already starting on multivendor interoperability testing. Those operators include KDDI, AT&T, Bharti Airtel, China Mobile, China Telecom, Deutsche Telekom, KT, NTT Docomo, Orange, SK Telecom, Softbank, Telefónica, TIM and Verizon.
The ORAN Alliance was founded by five operators – AT&T, China Mobile, Deutsche Telekom, NTT Docomo and Orange – and even secured Ericsson’s membership last month, although the Swedish vendor has not contributed code as yet, and may mainly interested in joining to keep a watching brief on a potential challenge to its model.
Network slicing starts with fixed networks and a cloud-native core
Network slicing, we believe, will be essential to the business case for 5G, for operators which are serious about expanding into new industrial markets. However, full 5G slicing is reliant on the MNOs implementing full cloud-native 5G cores and other capabilities which are absent from the first wave of roll-outs. But slicing is already emerging in the fixed-line area with FANS (fixed access network slicing) – not the fully dynamic, end-to-end platform of the 5G vision, but nevertheless a useful step in the right direction.
Nokia is the most willing of the big network vendors to risk conflicts with its operator customers by providing service platforms that support enterprises or MVNOs directly. So its interest in slicing is unsurprising, since this would enhance the ability to support many different customers from a single cloud core, while guaranteeing each one a fully isolated, secure and personally optimized virtual network.
The Finnish vendor believes its FANS technology will help customers to move more easily between fixed broadband operators, initially, and in future – when 5G enhancements allow the radio air interface itself to be sliced – converged slicing will allow for similar movement between MNOs.
It is starting to trial its fixed access network slicing in Lebanon with service provider Ogero, which says this will be a solution to “the long-lasting challenge of fixed infrastructure sharing”. In some cases, these challenges need to be met because national regulators are stipulating open access to an incumbent’s wireline network; in others, operators are interested in such changes to extend their business model or reduce costs through co-ownership. Old methods of implementing unbundling are too rigid to support the many variations and combinations of sharing, neutral host, wholesale and retail models that will be required in the fiber/5G era.
Nokia’s FANS enables operators like Ogero to “create several discrete network slices over existing fiber-to-the-home networks in a fully programmable way”, said the vendor. The solution is based on the supplier’s cloud-native software platform Altiplano and its software defined network (SDN) programmable Lightspan FX optical line terminals (OLTs). Ogero aims to maximize the utilization of its fiber infrastructure, and therefore its ROI, by providing virtual slices to a range of service providers, MVNOs, ISPs and even fellow MNOs.
Some vendors are coming at the challenge from the other direction, by virtualizing the customer premises equipment (CPE) so that it is open and able to support many applications and connections, regardless of provider. Earlier this month, Dell EMC and ADVA announced a universal open CPE initiative of this kind.
Sandra Motley, president of Nokia’s Fixed Networks Business Group, said: ” Network slicing gives service providers the right capabilities to innovate while unlocking the full business potential of fiber-to-the-home markets. Software-defined access networks will play a key role in revolutionizing conventional network infrastructure management models; helping to simplify operations and deliver better, lower cost services than previously possible.”
Rakuten adds NEC’s Massive MIMO to its cloud-native 5G architecture
Japan’s Rakuten was one of the most closely watched of the world’s operators, because of its disruptive, greenfield status, before it even got 5G spectrum. Now it has those airwaves, and before its modern, cloud-native 4G network is live, it is already discussing how it will go forward into 5G.
It is fascinating to the wider industry because it provides a glimpse of how other cloud providers could get into the mobile world, and how that might create havoc for established operators. Companies coming from the cloud industry clearly have the cloud-native technologies that underpin 5G at the heart of their operations, and so can leverage them from day one for scalability and flexible service delivery.
Chief information and security officer Anand Prasad recently said in an interview that, as an Internet player, Rakuten is using its IT experience to differentiate itself by taking a completely cloud-native approach, with plans to virtualise the radio network. He said the cloud has been used for a long time, and “if you’re using it for the mobile network, I think there isn’t a big challenge”.
By contrast, MNOs generally have a major organizational and cultural shift still to make to embrace cloud technologies fully and so make the most of 5G’s potential.
However, Rakuten does not have experience of many of the areas of network operations and monetization in which its rivals will be so experienced, so the success of the new entrant is by no means assured. The risk for NTT Docomo, KDDI and Softbank, however, is that their new competitor will assess ‘success’ very differently from them. While they are still investing huge sums to gain higher market share and ARPU, Rakuten will be looking at its 4G and future 5G network as a way to boost customer engagement and quality of service, and so shift more content, retail goods and web applications, including its financial services.
The additional revenue will come from these core activities, with mobile fees being a bonus rather than the lifeblood of their business. While most MNOs struggle, and often fail, to become content and web services providers, Rakuten has those operations already established, and so does not need the mobile network itself to be hugely profitable, as long as it signs up the users. If AWS ever does launch as a mobile operator, it will have the same motivations, and pose the same threat to established rivals – an entirely different set of assumptions about the value of connectivity, and the willingness to make 5G a loss-leader to improve ecommerce, content and finance revenues.
At this year’s Mobile World Congress, Rakuten announced a long list of suppliers of its 4G RAN and core, which will all be designed to be fully cloud-native and 5G-ready. Indeed, it promises to support in 4G a number of concepts which, for most MNOs, are still far from at-scale reality, even in early 5G. These include virtualized RAN, edge computing, network slicing, end-to-end automation supported by AI, an IPv6-based transport network, a unified OSS layer, and a cloud-native packet core with full control/user plane separation – all of which, if well implemented, will make Rakuten’s network flexible, cost-efficient and sliceable.
Suppliers include Nokia, Altiostar, Cisco, Intel, Red Hat, Tech Mahindra, OKI, Fujitsu, Ciena, Netcracker, Qualcomm, Mavenir, Quanta, Sercomm, Allot, Innoeye and Viavi. Now they are joined by NEC, in Rakuten’s first 5G-specific announcement, and its first since being allocated 5G spectrum by the Japanese regulator in April (see inset).
Rakuten and NEC are jointly developing a 3.7 GHz Massive MIMO 5G antenna radio unit to work with the fully virtualized RAN and core and support commercial 5G services from mid-2020. The new operator is already working with NEC’s software subsidiary Netcracker on its end-to-end BSS/OSS solution.
In the sub-6 GHz bands where Rakuten will launch LTE, it has architected its own custom solution in which antennas from Korean manufacturer KMW work with Nokia remote radio heads, which in turn connected by fiber fronthaul to distributed baseband units, located at 4,000 edge sites and running Altiostar’s vRAN software and Cisco’s NFVi. These are linked over midhaul connections to two central units running the most centralized vRAN functions. For 5G, the architecture will be the same, but the NEC Massive MIMO arrays will be added to support higher spectrum bands. Rakuten plans to deploy 16,000 of these units.
Rakuten Mobile’s CTO, Tareq Amin, said in a statement: “NEC’s technology and expertise will allow us to not only cost-effectively create a highly secure, high quality 5G network, but by designing and producing the antennas in the local market, we look forward to contributing to the development of the Japanese telecommunications industry and economy.”
Japan allocates 5G spectrum and sets targets:
In April 2019, Japan’s Ministry of Internal Affairs and Communications (MIC) assigned spectrum in the 3.7 GHz, 4.5 GHz and 28 GHz bands through a beauty contest to the four mobile operators in Japan.
Spectrum was assigned as follows:
NTT Docomo: 3.6-3.7 GHz, 4.5-4.6 GHz and 27.4-27.8 GHz
KDDI: 3.7-3.8 GHz, 4.0-4.1 GHz and 27.8-28.2 GHz
Softbank: 3.9-4.0 GHz and 29.1-29.5 GHz
Rakuten: 3.8-3.9 GHz and 27.0-27.4 GHz
The operators committed to the following coverage of the population five years after issuance of the licence:
NTT Docomo: >90%
The mobile operators also committed to the following investments in their 5G networks:
NTT Docomo: $7bn