Had Mobile World Congress gone ahead as planned this week, open RAN technologies would certainly have taken a central role. This topic has everything required to generate an MWC buzz – large operators banding together to change the industry; plucky start-ups taking on the giant vendors; the prospect of a major disruption of the industry.
Even without the giant meeting in Barcelona, many of the ingredients were present in offline announcements and debates. Huawei and Ericsson duly played their role as Goliaths, resisting the pressure to embrace open architectures. Challengers like Mavenir and Parallel Wireless announced important new partnerships and operator deals. Companies from the data center world such as Intel, Dell and VMware, eager to penetrate the mobile networks fortress, teamed up with smaller suppliers and the open alliances to try to prove the big OEMs wrong. And large operators such as Vodafone were keen to use all this to make it clear that, if their usual vendors did not deliver something new in terms of architectures and network economics, they could go elsewhere.
All of this has happened before, of course. WiMAX, Carrier WiFi, small cells – all of these have promised a new, open ecosystem which would encourage a diverse supply chain and affordable, multivendor networks. All have been used by large MNOs to pressurize their vendors with the threat of a fully open platform – Vodafone using its MWC keynote address to tell its suppliers to get a move on with 4G, or it would adopt WiMAX instead, was a classic example.
In reality, large operators have made very little real progress to an open supply chain and a multivendor network. A few have introduced new, disruptive players to the OSS/BSS and core layers, as these have gradually been virtualized, but the RAN has remained obstinately proprietary and in the hands of, actually, fewer suppliers than ever before. This is unsurprising – the RAN accounts for almost two-thirds of most operators’ network capex, which on one hand provides a motivation to adopt lower cost platforms, but on the other, makes it an investment of such importance that no risk can be taken with it.
Where there have been moves towards a more open platform, these have usually been in greenfield environments – ultra-rural deployments, private networks, generally based on small cells. These are important as demonstrators, but do not help to prove the thesis on which the promise of a new market and technology structure rests – that a virtualized, open, multivendor RAN, running on merchant processors, really could match or even exceed the performance and robustness of dedicated, integrated systems, with decades of engineering behind them.
Huawei and Ericsson say the premise is flawed (even though Ericsson is a member of the O-RAN Alliance). The two main open RAN industry groups, O-RAN Alliance and Telecom Infra Project (TIP), are attracting more and more operators and innovative vendors to contribute to their specifications and bolster their arguments in support of the motion. Importantly, they have announced a way to work far more closely together, which will help to bring together an emerging sector in danger of fragmentation.
On the software front, some of the proofs are starting to be made. Virtualized network functions (VNFs) for RAN, and open interfaces and protocols, are all emerging from large and small suppliers and being tested to destruction in commercial trials.
Some of the hardware, too, is starting to look convincing, especially for 5G. The new Evenstar project within TIP, which will develop an open remote radio unit (RRU), is a promising development, as are open 5G APIs from the Small Cell Forum.
The next frontier will be to persuade the industry that, in the macro layer rather than just in smaller cells, general purpose processors, common accelerators and Ethernet really can do the job of the proprietary platforms, and that they can do it at lower cost. Intel’s announcement of its first 5G base station system-on-chip is an important step forward (there will be full coverage of this in next week’s Wireless Watch). But having Intel repeat its dominance of the server market in network infrastructure would not deliver an open ecosystem. Greater access to investment funds for silicon innovators, at least outside China, would help the cause a great deal; meanwhile Marvell is a strong presence, and increasingly open in focus, while few would bet against another attempt by Qualcomm to re-enter the infrastructure space.