5G operators head for the clouds

Back in the 1980s computer giant IBM narrowly avoided being broken into parts in the same way that phone company Bell had been in the previous decade. Part of the reason for the US government abandoning its antitrust investigation of Big Blue was then president Ronald Reagan’s enthusiasm for the way the company had managed to break into the Japanese market at a time when electronics companies from the Far East were laying waste to American suppliers thanks to their willingness to sell products more cheaply.

In the event, IBM’s reprieve proved to be less effective than its management had hoped. At the start of 1993, the company won the record for the biggest ever corporate loss in US corporate history, at least up to that point. IBM’s weakness proved not to be antitrust lawyers concerned over the way it dominated the mainframe and minicomputer markets, leading to many would-be competitors quitting. The old adage was “no-one got fired for buying IBM”. The weakness was the personal computer the company’s own engineers had developed, which paved the way to far cheaper alternatives to its own, highly profitable proprietary hardware. By the end of that decade you needed a really good reason to buy IBM’s big-iron machines and not a network of PC-based servers.

Geopolitical concerns, again driven by US presidents, lie at the heart of a similar development that may topple an oligarchy of suppliers in the wireless communications business. At the March meeting of the US Federal Communications Commission, acting chairwoman Jessica Rosenworcel pointed to a “supply-chain challenge”.

In a statement prepared for the meeting, she stated, “We have only four major vendors for mobile network equipment to choose from, none of which hails from the United States. Plus, the vendors that have grown fastest in recent years are from China, in part because the Chinese government deploys powerful industrial policies to make their equipment cheaper to deploy than the alternatives.”

Fears over national security and competitive positions have combined to make China public enemy number one in telecom for the US, a policy that the world’s largest economy is successfully exporting to Europe and elsewhere. This in turn has led to the FCC deciding it’s time it got behind a movement that has been developing for several years among telecom operators and component suppliers.

There is a good reason why there is just a handful of suppliers of advanced wireless telecom gear: it is expensive and complex to develop and support. The radio-access network (RAN) of base stations, controllers and routers that funnel data in and out of the core fibre-optic network and onto the airwaves is based on an enormous set of standards documents. And they have to meet stringent requirements on uptime and response times, particularly for 5G, which is being promoted for applications such as networked robots that need to cut round-trip latency to the order of single milliseconds.

However, the OpenRAN movement recognises that 5G and its predecessors are open standards. It should be possible to devise ways to have bits of equipment come from different suppliers and still work together. Some of the functions potentially can be moved to cheap commodity computers that are programmed in the same way as cloud servers are today, so that if one breaks, the workload just moves to the nearest available node that runs the same services but not necessarily on identical hardware. In principle, this would improve uptime overall and work out cheaper, not least because much of the hardware winds up being cheaper, unbranded 'white box' versions. What operator is not going to like that option? The FCC is now keen on the idea, too.

“At the FCC, we are starting the first-ever inquiry into OpenRAN,” Rosenworcel noted. “Today, the RAN is the most restrictive and most expensive part of the network, in part because all of its major components have to come from the same vendor. There is no way to mix and match. But if we can unlock the RAN and diversify the equipment in this part of our networks, we may be able to increase security, reduce our exposure to any single foreign vendor, lower costs, and push the equipment market to where the United States is uniquely skilled – in software.”

One of the ironies of this exercise is that China Unicom coined the term OpenRAN as part of its proposal for opening up the RAN environment at the 2013 ACM Special Interest Group on Data Communication (SIGCOMM) conference. At the same 17 March meeting where the FCC said it was speeding up its involvement in OpenRAN, the agency was getting ready to ban China Unicom from the US market.

Across the Atlantic, having decided to wind down the reliance of network operators on Huawei hardware, the UK’s chancellor of the exchequer Rishi Sunak said last November the government would put £250m into diversifying 5G sources, in a programme following the decision to set up a task force headed by former BT CEO Ian Livingstone to look at OpenRAN’s potential. In January, operators Deutsche Telekom, Orange, Telefonica and Vodafone signed a memorandum of understanding to work together on a European OpenRAN ecosystem.

Although it now has the theoretical backing of large operators, it may be smaller service providers such as Japan-based Rakuten who do much of the early deployment. According to Rethink Technology Research’s RAN-research group, it is the alternative operators who are most likely to want to use OpenRAN in the early stages. Part of that is because the bigger suppliers are already well under way in their 5G roll-outs, which are by necessity based on the products sold by the big guns. Even so, growth in OpenRAN could be fast. According to Rethink, by 2026, OpenRAN deployments could account for almost 60 per cent of global RAN capital expenditure, which will total $32.3bn by then.

Despite the obvious attraction of OpenRAN, the impending death of the major suppliers may be exaggerated. In a webinar last week organised by chipmaker Analog Devices, which is among those now selling reference designs for the radio units, Raj Singh, executive vice president of Marvell’s processors business group, pointed out, “There are things you can virtualise and there are things that are more difficult to virtualise, especially as you get closer to the radio. As you disaggregate functions, that disaggregation introduces some level of latency. So you can’t disaggregate everything willy-nilly. You have to do it judiciously.”

One major headache in the push to virtualise is how to deal with technology such as massive MIMO, which is crucial to delivering high-speed data to multiple users in built-up areas. Ideally, you would want multiple closely spaced base stations to cooperate to try to maximise spectrum usage, but doing this with hardware from multiple vendors is going to be tough, whereas higher-level functions such as diverting packets between Wi-Fi and 5G nodes is likely to be easier to handle given all the latency constraints 5G has. However, even that is not going to be easy.

“The challenge for OpenRAN is going to be the unboxing experience. Service providers today are used to dealing with a one-stop shop. Dealing with a disaggregated network is going to be difficult. Unless the set-up is simple, it is difficult to imagine [OpenRAN] will be widely deployed,” said Jaydeep Ranade, director of wireless engineering at Facebook Connectivity.

Singh said it will call for suppliers and users to cooperate with each other much more, which is happening to a degree. For example, Vodafone has set up testing for OpenRAN and is now looking at ways to publish the results it obtains. “OpenRAN requires us to collaborate, often with people we may treat as competitors. And that is not just at the software level but levels we haven't envisaged yet,” Singh said.

Although cooperation will likely help get equipment to work together more easily, there is another outlet for today’s major suppliers, assuming they are allowed to sell into a country in the first place. “Someone has to put this together,” Singh said. “There is a role for system integrators. Even a role for incumbents to perform system integration and do the stitching together.”

When IBM suffered its fall from dominance in the early 1990s, it largely reinvented itself as a systems integrator, something that companies such as Ericsson and Nokia could do themselves – though the likes of Cisco are looking hungrily at the market, knowing that as a home-grown supplier they will be the popular choice for US regulators. Though it has serious challenges to face, most of the indicators point to OpenRAN getting an easier ride over the coming decade.

“It’s not a question of if OpenRAN is going to happen but more a question of exactly when,” says Paco Martin, head of network planning at Vodafone.