These are early stages for the dream of an open RAN in which key elements from multiple vendors can be combined, swapped and reconfigured without sacrificing any performance or interoperability. A few large trials may be mainly meant as warning signals to existing vendors to take a more open, cost-effective approach, rather than a real effort to lower barriers to newer, smaller suppliers. The need to implement 5G quickly and robustly in some markets may drive some operators, albeit reluctantly, to stick with the certainty of an end-to-end solution from a well-established supplier with decades of R&D behind their offerings.
But those objections were raised years ago in the data center, yet open source and virtualization took hold there and became the norm. The same could happen in the RAN, and in the wider end-to-end mobile network, if sufficient operators join the cause, and refuse to tolerate the vendors making incompatible implementations of the new interfaces, as they have with the old ones like CPRI and X2.
This is the significance of efforts by NTT Docomo, Vodafone and others to light a fire under emerging open specifications. Docomo’s Tokyo network, based on ORAN Alliance interfaces, and Vodafone’s new trial, using Telecom Infra Project (TIP) OpenRAN specs, are important because they take these architectures out of the realms of lab tests, proof of concepts and demonstrations, and into live networks where they can prove their capabilities (or not). Docomo’s launch was of a pre-commercial 5G ORAN system; Vodafone has announced field trials in multiple countries.
If they prove themselves, there will be real hope that operators can get away from a world where, even though standards are set, vendor have the market power to implement those standards in their own, incompatible ways, and get away with it because there is so little choice in the market.
Earlier this year, the UK government’s Telecoms Supply Chain Review Report summed up the problem, commenting on the paucity of RAN suppliers, and commenting: “Greater interoperability and more open interfaces will be required to facilitate new entrants. It is not sufficient that interoperability is included in technical standards – industry must work to ensure equipment from different vendors is interoperable in real world deployments.”
Of course, such statements by government, and operators’ moves towards open networks, are often taking place against the backdrop of the uncertainty over Huawei’s status as a 5G supplier, even when its name is not mentioned. Vodafone is a major customer of Huawei in several markets, including the UK, but Huawei has so far steered clear of the open RAN groups, even though Nokia and Samsung are active in TIP and ORAN, and Ericsson has also joined ORAN (though has not been very active, reportedly).
Vodafone acknowledges on its website that “the global supply of telecom network equipment has become concentrated in a small handful of companies over the past few years. More choice of suppliers will safeguard the delivery of services to all mobile customers, increase flexibility and innovation and, crucially, can help address some of the cost challenges that are holding back the delivery of internet services to rural communities and remote places across the world.”
Nick Read, CEO of the Vodafone Group, said in a statement: “We are pleased with trials of OpenRAN and are ready to fast track it into Europe as we seek to actively expand our vendor ecosystem. OpenRAN improves the network economics enabling us to reach more people in rural communities and that supports our goal to build digital societies in which no-one is left behind.”
Details of the new trial are sparse but it will take place first in the UK, where Vodafone will test technology from Mavenir and Parallel Wireless in about 100 rural locations which are underserved by cellular. It will also use its own open-sourced CrowdCell small cell, with RF technology from Lime Microsystems (whose open platform has already been used in rural-focused systems by EE).
This builds on work that was kicked off in June 2018, when Vodafone and Telefónica issued a joint request for information (RFI) on systems that would comply with TIP OpenRAN specs and requirements. At the TIP Summit in London in October last year, they announced the most compliant systems, all of them from small suppliers (see inset), and announced that they would start on limited trials of some of the technologies.
In Vodafone’s case, it followed up on the RFI in two ways. It selected two vendors to turn its CrowdCell small cell design, which it has open sourced, into a commercial product. Those partners are Lime Networks on the RF side, and Alpha. And it initiated three new OpenRAN trials, working with Parallel Wireless in Turkey, and with Mavenir and Altiostar in Democratic Republic of Congo and Mozambique. Open CrowdCell was one element of these trials too. It has also carried out lab tests with its South African subsidiary, Vodacom.
Telefónica, for its part, has been working with Altiostar, Mavenir and Parallel on trials in Peru, Colombia and another unannounced Latin American market, focusing on improving return on investment in rural networks.
There is still a long way to go to build a global ecosystem on these early efforts. A major enabler of a fully open RAN will be harmonization of the key specs, and so far this is not on the horizon.
There are three major initiatives devising interfaces and open platforms for the RAN (and other network elements like core and transport, in some cases). These are TIP, the ORAN Alliance and the Open Networking Foundation’s (ONF’s) CORD (Central Office Re-architected as a Datacenter) project. Vodafone says that TIP OpenRAN is fully compatible with ORAN Alliance specifications and that the two initiatives are “complementary”, but neither group has offered real details of the level of that compatibility or, importantly, whether it requires integration effort by the MNO.
TIP’s OpenRAN working group has prominent support from European operators like Vodafone, Telefónica and Deutsche Telekom. As well as this group, TIP has other related groups focused on enablers like virtualization and automation, and is promoting some reference designs, such as the OpenCellular base station, to make it cheaper and easier for new vendors to adopt common platforms and design mobile gear.
The ORAN Alliance was initially based on xRAN code contributed by AT&T and by China-driven Cloud-RAN Alliance, though its first commercial deployer is Docomo. CORD is the longest established group, and is also operator-driven, though its initial focus was more weighted towards the core.
It will be important that these different initiatives find common ground or the industry will remain fragmented and incompatibility – with its accompanying costs and complexities – will remain, between different interfaces rather than different OEMs. the success of the open RAN vision is likely to rely on a measure of convergence of the open source efforts in the near future – unless one of them clearly becomes the industry’s de facto standard by sheer weight of adoption.
Last year, there were indications from Deutsche Telekom’s Axel Clauberg – chair of TIP and also a leading light in several other open groups – that TIP and ORAN could converge their efforts, though that has not happened. He said a year ago that Telefónica and Vodafone, as well as being leaders of the TIP Open RAN project, plan to base their work on the ORAN specifications. That, he intimated, would lead to coordination of efforts, though not a merger of the groups.
“They have not done any specs from the ground up. They have not built architecture but are looking at how the architecture specified by the ORAN Alliance could be built and deployed in a commercial way,” he said. “In this area, TIP would do solutions integration and prove it is possible to build such an open RAN in real life.”
In another sign of crossover between different open groups, Radisys, which is a leading light in CORD, and is now owned by Reliance Jio, has been using TIP’s Open RAN framework, interfacing its 4G baseband software running on an x86 common platform, with a remote radio head from Baicells, using the TIP specification.
Whichever groups continue to lead the way, there are key elements of an open RAN which will be essential to support:
- Consistent, open interfaces between key elements such as radio head and centralised baseband processor; between RAN and packet core; or between different base stations in one network.
- Virtualization of some or all of the RAN baseband functions on commoditized or white box hardware, with flexibility to change the split between centralized and site-based functions according to the use case.
- A software-defined radio (SDR) that can be reprogrammed to support new combinations of spectrum bands or radio technologies.
- Software-defined networking (SDN) control and open management and orchestration (MANO), to allow virtualized network resources to be assigned flexibly where required.
These are the key elements which will deliver the significant 5G deployment cost savings which open network proponents are promising. Rethink’s own models indicate an average of 35% difference in five-year total cost of ownership between a single-vendor RAN and an open, multivendor RAN, assuming both are virtualized (the difference is closer to 50% when comparing a conventional RAN with a virtualized, disaggregated open RAN). Parallel Wireless says deployment costs (mainly capex) will fall by 30% between now and 2022 even in traditional networks, but that figure will be 50% if operators move to open networks.
But the cost savings will be irrelevant if the networks do not deliver the performance, security and scalability of the conventional RANs. That is why all eyes will be on these early trials of open systems, because proof points are badly needed that open networks really can perform as well as the platforms that have been developed over decades by the large OEMs – and that they can do it soon, not after years of evolution or millions of dollars of operators’ integration dollars to get them to the right place.
So far, OpenRAN trials are heavily focused on small cells and rural areas – both areas where the major OEMs have failed to provide equipment at viable price points for at-scale, profitable deployment. But with operators moving to 5G at an early stage in many countries, their focus for the new generation is mainly on macrocells and urban areas. It is not yet clear how the open architectures will translate to the macro network, or whether there is any chance of displacing the big vendors from this layer.
They may be forced by operators’ collective pressure to adopt fully open interfaces between their radio heads and basebands, and between different base stations, and that would achieve interoperability between their respective systems. But that does nothing to usher new, smaller suppliers into the macro layer, and with the first wave of 5G roll-outs underway, it seems unlikely this will happen until a future generation of 5G.
When asked recently why it had not joined any of the open RAN initiatives, a Huawei spokesperson said that open RAN architectures were a poor substitute for dedicated equipment. Of course the comment was self-serving, but until, and unless, early trials prove the capability of the new platforms, it would be hard to prove him wrong.
Mavenir opens vRAN lab in Ericsson’s heartland:
Mavenir has been an early player in disaggregated, virtualized cores and through acquisition, extended its reach into the vRAN too. It recently opened its second
Center of Innovation, in Stockholm, Sweden. This will focus on Mavenir’s development of open vRAN systems based on ORAN and OpenRAN interfaces, as well as on 5G and IoT.
The center will be led by Mikael Rylander, SVP and general manager of Mavenir’s RAN business. Clearly, locating it in Ericsson’s homeland will help with securing relevant skills, and indeed, Mavenir’s new general manager for emerging business, Ani Basu, was formerly with Ericsson and is based in the Stockholm office.
Mavenir’s other innovation center is in Dallas, Texas, at its headquarters.
To mark the opening of the Stockholm center, Mavenir demonstrated the transmission of data traffic over an ORAN interface in a 5G New Radio (NR) network. The transmission at the physical layer was based on ORAN Split 7.2 between a distributed unit implemented on an x86 server, and a simulated remote radio unit.
Mavenir also published a piece of research, conducted by Senza Fili, which said 43% of US service providers are considering replacing their current vendors in order to meet cost reduction targets of at least 25% in the 5G era.
The most compliant vendors with TIP’s 2018 RFI:
Four months after the Telecom Infra Project’s OpenRAN project group issued a request for information (RFI) in June 2018, two of the MNOs which adopted it, Vodafone and Telefónica, announced the vendors which best met the scope of that RFI. The aim of the process was to identify and highlight suppliers which might be well positioned to execute on the vision of an open, disaggregated, virtualized mobile network in future.
In future, we might expect these independent suppliers to be joined by some new challengers, currently being incubated under the TIP Enterprise Acceleration Center (TEAC) program. Companies like Athonet and BISDN, among others, are being supported by four TEACs, established by BT, DT, Telefónica and SK Telecom, to work on TIP-focused solutions.
Those recognized by OpenRAN and the two operators last year were:
- For most compliant end-to-end platform: Mavenir; Parallel Wireless; Altiostar.
- For 2G software: Fairwaves; Parallel Wireless.
- For 3G software: Mavenir; Parallel Wireless.
- For 4G software: Altiostar; Parallel Wireless; Radisys.
- For remote radio head/ radio hardware: Baicells; NEC; Parallel Wireless.
- Innovators: ASOCS; Phluido.
- Challengers: Comba; Dali; Fairwaves; Vanu.
This activity has brought TIP, or at least its software-based RAN efforts, a big step closer to commercial reality, and will be a big profile boost to the selected companies.
Many have been pioneers in key areas of the emerging, alternative wireless architecture. Altiostar was early to recognize the need for a vRAN with Ethernet fronthaul to avoid the cost and lock-in of CPRI; Radisys has been a leading light in another important open initiative, CORD (part of the Open Networking Foundation); Mavenir was one of the first non-OEM providers of a virtualized and disaggregated evolved packet core (EPC); ASOCS has a pioneering vRAN chipset. Others, like Parallel and Baicells, have been firmly focused on small cells and densification. The only big OEM to feature in the list is NEC, but of course, in the RAN, the Japanese firm is a minnow.