Your browser is not supported. Please update it.

12 February 2016

C-RAN shrinks and finds its sweet spot in the enterprise

At last year’s Mobile World Congress, there were some interesting demonstrations of Cloud-RAN, from the groundbreakers like China Mobile and Telefonica, but a general assumption that this was a technology for the future, and virtualization would be more immediately relevant, to the vast majority of operators, in the core or CPE than the RAN.

This year, there will still be those high-concept demoes, but there is also set to be a far more short term, pragmatic approach to Cloud-RAN. This reflects that the architecture is ticking boxes for a wider range of MNOs, with shorter timescales in mind, than originally expected. But for most, it will not be about re-architecting the wide area macro network around a whole new approach – the risks are high, many technology elements unproven, and most carriers are still in the midst of rolling out conventional LTE networks. Instead, they are looking at adapting the approach – which virtualizes baseband processing on a commodity box shared between many cell sites – to smaller scale deployments, notably enterprises and other scenarios where dense local zones of capacity and coverage are increasingly required.

That means changing the technology assumptions. Even the term ‘Cloud-RAN’ is not necessarily the right one, since many of these more localized architectures rely on shared controllers or servers, but not on the cloud. ‘Centralized-RAN’ is heavily used, and that can include fairly established approaches to sharing resources (remote radio heads with a baseband hotel), and not necessarily the cloud or even NFV, the emerging standard for virtualization.

And an enterprise C-RAN is not about macrocells and long CPRI fronthaul links over fiber between radio/antennas and cloud-based servers. It is only practical if more modest components can be made to work with the same centralized, shared architecture, to achieve the promised benefits of C-RAN – flexible allocation of resources, lower cost of ownership, smaller and greener cell sites, and the ability to support multiple use cases or service providers on one ‘sliced’ infrastructure.

So much of the most real world work is being done by the Small Cell Forum and its members, and companies like CommScope, Spidercloud and ip.access already offer centralized RAN solutions for enterprises and venues, which will evolve towards full virtualization in future. The Forum itself is working in an interface, NFAPI, which will be a critical enabler of a standardized approach, allowing the baseband, radio and antenna (all integrated in a traditional small cell) to be split at various points, and for the elements to be mixed and matched.

There has been significant activity in recent weeks in the small cell side of C-RAN. Spidercloud, whose big news in Barcelona last year was that Cisco would sell its system, has now announced that it will be part of Verizon’s trials of LTE-Unlicensed, whose initial target market is likely to be the indoor enterprise network. CommScope acquired Airvana last year, adding that firm’s OneCell C-RAN enterprise solution to its DAS-oriented in-building portfolio, and will be outlining the product roadmap at MWC.

IP.access, one of the veterans of small cells, has announced Viper (Virtualized, In-Premises Enterprise RAN), which integrates the company’s 3G and 4G access points with virtualized access control gateways. This can also be deployed with SCaaS (small cells as a service) options to support a fully managed network for an enterprise – an increasingly common approach in corporate WLANs, and an essential enabler of similar uptake of cellular in-building technology.

Viper aims to make it as easy to deploy small cells as WiFi, by providing the operator or integrator (or the enterprise itself) with in-built security, automated provisioning and network configuration, and self-organizing network capabilities. The access points require only Ethernet and mains power connections.

Viper also features ip.access’ SUMO multi-operator capability, which allows one small cell system to support users from any mobile network. Lack of multi-operator support has been a significant barrier to enterprise uptake of small cells, since they need to provide coverage for employees and visitors, regardless of their cellphone contract, and this issue has become more pressing with the rise of BYOD (bring your own device) workplaces. Implementing multi-operator technologies like the 3GPP’s MOCN avoids the need to roll out more than one cell per location, and ticks another box when comparing with WiFi, with its inherently multi-operator approach.

“We see small enterprises as a key market for this technology. Over 95% of businesses have fewer than 100 employees, and many of them are located in office buildings with poor mobile coverage. The Viper platform allows everyone to get five-bar 3G and 4G coverage anywhere in their premises as easily as deploying enterprise WiFi,” said CEO Malcolm Gordon.
The access points themselves come in a variety of sizes to suit large and small companies, and support many spectrum bands, including the US’s upcoming ‘Citizen’s Broadband’ in 3.5 GHz, and licensed shared access approaches. The virtualized gateways can be deployed on servers or in the cloud. They provide a single, standard interface to the core and OSS/BSS for all the small cells, which are treated as a single AP, with small cell signalling offloaded to the gateway.

Even outdoor-focused Parallel Wireless is eyeing the indoor enterprise. Parallel has been a pioneer of adapting C-RAN approaches for small cell and localized deployments. Its focus has been on bringing coverage cost-effectively to rural areas – as adopted by UK mobile operator EE, its solution covers a village using a zone of small cells, which are attached to a single controller, so that only one backhaul link is required and most traffic can be handled locally. This greatly eases the cost and load of the backhaul – often a sticking point in rural coverage.

Now, like ip.access, it has adopted the mantra of making indoor small cells as easy to deploy as WiFi for building owners and enterprises. President and CTO Rajesh Mishra said in an interview with ThinkSmallCell: “In future, I would expect more and more building owners and businesses to arrange and pay for installation of their own cellular equipment for connection to the network.

I foresee that the emerging ecosystem to enable this will consist of many partners specializing in their own area – ODM equipment vendors, system installers/integrators and outsourcing companies.”

To address this broader ecosystem, Parallel plans to produce a white box reference design to help low cost manufacturers to mass produce small cells which can interwork with those from other suppliers, and with DAS. The firm says it has tested interoperability between its HetNet Gateway, the cloud-based controller for its rural C-RANs, with 15 eNodeB vendors and about the same number of packet core suppliers. The Gateway supports MOCN and can therefore enable a managed service, whether the enterprise is using access points based on Parallel’s reference design (which supports 3G, 4G and WiFi), or those from a third party.

Unlike full Cloud-RAN, baseband processing is carried out at the edge in the small cells, so only data and signalling needs to be carried to the cloud, reducing backhaul overhead. This is an approach echoed by Quortus, a specialist in virtualizing the packet core on a small cell or local gateway, or in the cloud. Like Parallel, Quortus can therefore enable hosted enterprise services, as well as coverage in remote areas.

Small cell C-RANs are not just for the indoor enterprise. Ericsson and Dutch telco KPN have launched a C-RAN which coordinates access points built into bus stops around Amsterdam, providing hotspots of cellular capacity for users waiting for buses. The companies have worked with billboard operator JC Decaux, which has previously worked with Vodafone to incorporate small cells into its hoardings. The system uses Ericsson’s new Radio 2203 small cells, which are coordinated, and integrated with the macro network, using a shared baseband. Inter-site carrier aggregation is supported between macrocells in 800 MHz and the 1.8 GHz small cells.

Another important requirement for mass adoption of C-RAN architectures will be to reduce the reliance on fiber. In these smaller indoor roll-outs, distances and bandwidth are far smaller than in China Mobile’s Cloud-RAN concept, and so existing Ethernet connections can be used. Over larger areas, the high data rate and low latency requirements of C-RAN fronthaul have made fiber almost essential, and proved the biggest cost obstacle outside of markets (like Japan) where dark fiber is relatively cheap and plentiful. Dark fiber will be an important catalyst for wide area macro C-RANs, but for these more localized enterprise and metro systems (which could, in time, be aggregated to the cloud), wireless and even copper options are also gaining credence.

Of course, work is still proceeding on wide area C-RANs with their demanding fronthaul requirements. ZTE announced a new OTN (optical transport network) product optimized for fronthaul last week (see Wireless Watch February 4 2016), claiming to address “the growing challenges and pain points that today’s mobile operators face with CPRI”.

Nokia followed this week. Its solution supports wavelength division multiplexing, which increases fiber capacity by combining multiple wavelengths on one fiber. The company claims WDM reduces capex by up to one-third compared to dark fiber. Its 1830 Versatile WDM Module promises to support all data rates for CPRI and OBSAI connections, and to come with end-to-end operations, administration and maintenance tools, including active network management for passive optical systems.
The new offering joins Nokia’s broader mobile fronthaul portfolio, which also includes the 1830 VWM Photonic Managed Unit, an optical multiplexing product with management capabilities to support demarcation and SLAs for multiple providers; plus further OAM options and a fully hardened unit, combining multiplexing, colorizing and management functions for small cells.

Dimitris Mavrakis, principal analyst for intelligent networks at Ovum, said in the Nokia statement: “Radio network densification is one of the structural evolutions operators are now assessing on the journey to 5G, and while centralized RAN is a key element of this, mobile fronthaul is a key challenge. We are estimating that global spending on mobile fronthaul equipment will grow annually by 24% to $1bn by 2020.”

And BT has been testing the use of souped-up copper technology, namely G.fast, as an alternative to fiber in C-RAN. In an experiment conducted at its famous Adastral Park Lab, the UK telco demonstrated fronthaul speeds of 150-200Mbps over copper, claiming this would greatly lower the cost of deployment. Chip vendor Cavium, which has been developing processor platforms for C-RAN servers, was BT’s partner in the project, providing its Octeon Fusion–M base station and its ThunderX server processor technology.

“Using G.fast to deliver a cellular network is an exciting breakthrough for C-RAN and yet another world first for our team of researchers at Adastral Park,” said Tim Whitley, head of research and innovation at BT. “These technologies will play a key role in 4G networks and will be fundamental to 5G architectures. The trials are another step towards a fixed and mobile network which will support customers’ increasing demands for data.”

There were no details on actual costs or speeds, nor on distance, which is a challenge for current iterations of G.fast and might confine copper-based C-RANs to relatively localized links.

capacity.

This is a solution confined, for now, to rarefied carriers which are prepared to upgrade their networks and can also source devices which support 256QAM and CA (those will not be available until later this year, though there will be some improvement in performance even for handsets from Release 10 and above).

The first MNO to engage in a public demonstration of 1-Gigabit LTE (before it was commercialized and given its new name) was Australia’s Telstra, always a RAN frontrunner. In November, it worked with Ericsson on the first trial of five-carrier aggregation, combining 100 MHz across five separate 4G channels integrated on the carrier’s LTE network.

Telstra’s group managing director for networks, Mike Wright, wrote in a blog post that end-to-end tests achieved download speeds of over 950Mbps using a specialized field test application, as well as speeds of around 843Mbps over the internet to speedtest.net. “Many may be aware of the debate some year ago around the technical definition of what 4G should be, with a widely accepted view that 4G should achieve peak speeds in the range of 1 Gbps,” he said. “In the end the global consensus was that the term 4G could be applied to any LTE technology regardless of peak speed capability, but with LTE now reaching 1Gbps speeds in reality, LTE technology could be argued to be finally moving beyond the 4G barrier.”

Huawei recently announced a similar milestone in a lab trial with Singapore operator M1, which achieved gigabit download speeds on an LTE-A network by combining 4×4 MIMO; dual-band uplink carrier aggregation; 3CC triband downlink CA; and 256QAM. The companies used commercially available hardware and a prototype Category 14 Huawei device for the trial.

Ericsson also announced its new elastic RAN cloud architecture, which allows all its Radio System LTE baseband units to harmonize with adjacent units, within any topology – distributed, centralized or hybrid. Ethernet transport networks interconnect the basebands to form a single coordination area, an approach which Ericsson says will be “hyper-scalable”, while carrier aggregation and CoMP (coordinated multipoint) can further improve coverage and performance throughout the zone.

Meanwhile, ZTE says its theme for MWC is ‘The Enabler of Mobile Internet’. Like its rivals, it has two strands to its 5G roadmap – driving and later adopting emerging standards, and applying likely ‘5G’ technologies to current LTE platforms to alleviate network pressures in the shorter term.

ZTE has been calling its most advanced RAN portfolio ‘Pre5G’ for at least a year now, and much of its focus in 2015 was on massive MIMO, in which it has been a pioneer. Now it is adding ultra-density and multiuser shared access to the list of technologies it believes to be central to 5G and pre-5G, and will major on them in Barcelona. It is also promising news of 5G-oriented partnerships and tests with tier one operators from Europe, Japan, Korea and Russia, and with chip suppliers.

Like its competitors, ZTE will complement its RAN advances with progress in virtualization and software-defined networking (SDN). It will be launching its ElasticNet bearer network solution, which will target one of the most interesting aspects of virtualization – the ability for carriers to offer platform-as-a-service and dynamic MVNO capabilities by slicing their networks.