Suppose you’re a major supplier of fertilizer to the agriculture market and you have ambitions of digitally transforming your business. Your idea is to place thousands of sensors in 10-foot-by-10-foot blocks across the fields your customers farm. The sensors will continually report on soil and atmospheric conditions, enabling customers to adjust their fertilizer allocations to maximize yields. Meanwhile, your company gathers valuable data that it can use to target sales to other clients.
Such a scenario is possible today, but the cost is beyond the means of all but the largest agriculture conglomerates. The only way to build a network with the necessary reliability would be to contract with one of the handful of wireless service providers that dominate the U.S. market. Even then, your company would have to negotiate the details of how such a network would be operated, what level of control you would have and who would own the data.
The cost equation may be about to change, however. The next generation of wireless cellular networks – called 5G — for the first time will enable organizations to build their own networks and run applications on top of them. An open standard for radio access networks or RANs could bring the costs of such a project within reach of many more businesses.
The Open RAN specification, which the O-RAN Alliance introduced in 2018 and now maintains, is a reference architecture for a set of interoperable hardware, software and interfaces that can be built from off-the-shelf hardware. Supporters say it could have the same impact on the telecommunications industry that Unix and TCP/IP had on the data center in the 1990s when open standards disrupted the proprietary – and highly profitable – domain of a few large equipment makers, brought prices crashing down and changed the economics of data processing.
The Alliance will be out in force at MWC 2023, the former Mobile World Congress event, in Barcelona this week with an assortment of technical sessions and addresses by some major wireless carriers and businesses building compatible products. (SiliconANGLE and its mobile video studio theCUBE will be onsite to report and analyze the news and interview top executives and experts.)
A RAN is the radio element of a cellular network. It links wireless devices to transceivers and ultimately to the core network that connects to the internet. RANs are typically installed in each of the cells that make up a cellular network.
It’s a specialized piece of hardware that’s built to accommodate the speed and volume needs that are unique to wireless endpoints. “A RAN is a real-time application, so the levels of latency that are tolerable are significantly lower than what we see in other parts of the network,” said Sandro Tavares, director of product marketing for telecom business at Dell Technologies Inc. “It’s also extremely compute-intensive.”
Few can play
The need for speed has meant that RANs have always required custom silicon. Only a few large companies have the means to furnish that, which is why the market is dominated by just a few large players, including Huawei Technologies Co. Ltd., Samsung Electronics Co. Ltd., Cisco Systems Inc., LM Ericsson Telefon AB, Nokia Corp., Qualcomm Inc., Nippon Telegraph and Telephone Corp. and NEC Corp. Designs are proprietary and components are tightly coupled for stability and reliability.
The closed nature of the RAN market has inhibited innovation at the software level because each equipment provider controls its own operating stack. It has also kept prices high. By some estimates, the cost of deploying a RAN for a single mobile operator in a large metropolitan area can run to hundreds of millions of dollars.
Telecom carriers have been willing to put up with the expense because reliability is so critical to their operations, but Open RAN promises to fling open the doors to competition by making it possible for anyone to build the antennas, radios and baseband units that make up a RAN as well as software that runs on top of it.
“The current environment is stuck with a few major vendors and anything operators want to do has to rely on the development cycle of the vendor,” said Jai Thattil, head of service provider strategic marketing at Juniper Networks Inc., which is a vigorous proponent of Open RAN. “When you open up the market, more players will come in, smaller vendors can build new applications and deploy in days, time to market is quicker and you can pick and choose what you want.”
Systems makers rush in
The opportunity to get in on the market for telecommunications gear has caught the attention of vendors that used to be all but locked out of the market. Dell created a business unit in 2021 that is “aligning with the industry’s whole transition to open technologies,” Tavares said. “We expect our telecom business to grow significantly.”
Last week the company launched a line of bundled hardware, software and subscription services tuned to workloads at the edge of a wireless network as well as versions of its PowerEdge servers designed specifically for Open RAN deployment.
Hewlett Packard Enterprise Co.’s ProLiant DL110 server was built specifically to handle RAN workloads. It’s optimized to handle narrowband workloads with low power requirements and has expansion space for running computing workloads at the edge. Last week HPE double down on its enterprise wireless network business with the acquisition of private cellular network technology provider Athonet S.r.l. “We are the only ones that have deployed upwards of 10,000 sites of Open and virtual RAN,” said Geetha Ram, worldwide head of telco infrastructure at HPE.
IBM Corp. and its Red Hat subsidiary are involved in numerous Open RAN deployments and both Intel Corp. and Advanced Micro Devices Inc. have operations dedicated to the business.
Cloud platform providers, which some people have suggested could compete with wireless carriers in the future, are so far satisfied to play a supporting role. Amazon Web Services Inc. is pitching itself as an ideal platform for applications built on top of Open RAN. Microsoft Corp. is building analytics and control technologies for virtualized RANs and both it and Google LLC are members of the O-Ran Alliance.
Despite all the excitement, a variety of technical and competitive factors have combined to limit commercial deployments to just a handful. Only one carrier — Japan’s Rakuten Group Inc. — has bet its business on Open RAN, with more than 300,000 cells currently deployed. In the U.S. Dish Network Corp. is betting big on the technology but the cellular wireless leaders have so far only dabbled with tests.
For enterprises, Open RAN could significantly reduce the cost of building private 5G wireless networks, opening up new applications that are impractical today because of cost and complexity. By essentially disaggregating the elements of the base station, the standard could permit enterprises to build networks from components that precisely meet their needs as well as buy and build applications on top of them.
Dell’Oro Group Inc. expects the private wireless RAN market to grow at a 24% annual compound rate through 2027. Although the small-cell RAN market would only be a relatively tiny $1 billion at that time, “private wireless is a massive opportunity,” said Stefan Pongratz, a Dell’Oro vice president. “We believe there is a very large opportunity to address the limitations of enterprise Wi-Fi and mobile networks.”
Dell’Oro recently revised its Open RAN forecast upward, saying it expects the standard to account for between 15% and 20% of global RAN deployments by 2027. ABI Research Inc. is more optimistic. It predicts that shipments of Open RAN equipment will exceed 40% of total RAN equipment deployed in 2026.
“Vertically integrated equipment doesn’t have the ability to meet the fragmented requirements of the enterprise,” said Malik Saadi, ABI Research’s head of strategic technology. “You need to scale down or disaggregate. You need to mix and match components in line with the requirements of the segment.”
Although opening up one of the last proprietary bastions of information technology infrastructure sounds good in theory, the practical issues of transitioning to a new architecture are complex, particularly in a business that breathes reliability and performance.
The Open RAN specification has been praised for its scope, but it is only five years old and is still evolving. “It’s not lack of demand that’s creating the problem but the maturity of the standard,” said Saadi. “Specifications create bugs and those have to be fixed in subsequent releases,” he said, noting that about 40% of the new features in release 16 of the 5G specification fixed bugs from the previous version.
The specification will inevitably mature over time, but the risk-averse telecom industry is likely to move slowly until Open RAN equipment has been battle-tested. “They are preparing but it’s a slow process,” said Juniper’s Thattil.
Security is as much an issue as it is for any open standard. The Open RAN spec “hasn’t been designed for security from the start,” said Sylvain Fabre, a Gartner Inc. senior director and analyst. “I’ve got all these open interfaces that have to authenticate to each other.” A 2021 study commissioned by the German Federal Office for Information Security concluded that the specification at that time ”contains significant security risks” and “has not yet been adequately specified according to security-by-design” principles.
Proponents counter that the standard is, by definition, more secure because it permits a wider range of protections to be employed for functions such as threat detection, encryption, signature verification and zero trust administration.
The current case for return on investment is also shaky. The market of third-party equipment is still small and economies of scale have yet to kick in. There are also questions about the cost of integrating equipment and software from diverse suppliers. “Enterprise clients are seeing that the cost of going to a traditional operator is still cheaper than building it yourself,” said Raj Shah, managing partner at digital consultancy Publicis Sapient, the digital transformation consultancy arm of Publicis Groupe SA.
“Disaggregation requires standardization of a number of interfaces,” including those between RAN components such as radio units, distributed units, centralized units and the RAN intelligent controller, Saadi said. “These are still largely not standardized, although the O-RAN Alliance is working on that.”
The bottom line: There are few compelling reasons for carriers or enterprises to make big investments in the still-fledgling technology on top of their capital expenditures on existing equipment. “Carriers are very protective of their network,” said Randy Cox, head of product management for intelligent cloud at Wind River Systems Inc. “They want the highest-performing network and they don’t want just anybody to touch it. The downside of disaggregation is that it adds complexities in integration.”
The specification’s slow start has created a bit of a vicious cycle. “There was an expectation a lot of new players would enter the market,” said Vinay Ravuri, chief executive of EdgeQ Inc., which makes a chipset that fits within the Open RAN operating stack. “We’ve seen some but many have disappeared because funding dried up or they didn’t see enough uptake. There wasn’t enough silicon underneath to bring the cost structure down.”
That has made cost justification a dodgy issue for a technology that “requires a massive amount of investment to address all the various RAN segments,” said Dell’Oro’s Pongratz.
“Once it’s up and running it’s very cost-effective, but when you still have your legacy way of doing things, it’s going to cost you more,” said Gartner’s Fabre.
In theory, the total cost of ownership should decline as more players enter the market. “With a disaggregated network, buyers can pick and choose what vendors they want to work with and that drives lower TCO,” said Wind River’s Cox. “It also drives innovation with multiple players contributing to this disaggregated network.”
Strand Consult is one of the skeptics. The Copenhagen-based consultancy, which claims 170 mobile operators as clients, predicts that the Open RAN installed base will be below 3% in 2030. It says the economies of scale afforded by existing standards largely blunt any cost-saving benefits and that the benefits carriers could realize from adopting the standard would amount to savings of about 30 cents per month per subscriber.
Strand also accuses the O-RAN Alliance of opacity. “In general, there is little literature on Open RAN that can be classified as empirical, scientific, or peer-reviewed,” the company wrote in a report titled “Debunking 25 Myths of OpenRAN. “Outside of a few exceptions, most materials are policy advocacy promoted by OpenRAN providers.”
Perhaps the most daunting challenge Open RAN faces is the reluctance of entrenched equipment providers to embrace it. Although every major RAN maker has joined the O-RAN Alliance in at least a contributing role, their level of commitment varies widely, with some of the most entrenched players participating only in name.
“Companies like Nokia, Ericsson and Huawei haven’t embraced it because their own bacon is on the line and they don’t want to cannibalize their existing business,” said EdgeQ’s Ravuri.
Changes in attitude
Attitudes may be beginning to change, however. When HPE’s Ram raised the topic of open standards with NTT DoCoMo representatives at Mobile World Congress in 2019, “they politely told me I was full of it, that Open RAN will never take off,” she said. NTT has since done an about-face and now positions itself as an integration hub for equipment based on the specification.
European operators, in particular, are facing government-imposed deadlines to strip technology made by Huawei from their networks. “These operators have come to realize that the process of swapping Huawei equipment for alternative gear is costly, challenging, and could compromise overall performance,” Saadi said. “Such a situation will be unlikely if mobile networks are built with modularity in mind,” as is the case with Open RAN.
The chief technology officer of Ericsson, which has been guarded in its comments about the standard in the past, recently said that Open RAN will be in 40% of radio systems by 2030.
Qualcomm and Vodafone Group plc last fall said they will partner to develop 5G equipment with massive multiple-input multiple-output capabilities. MIMO is a wireless technology that combines hundreds or thousands of antennas in a single array to improve coverage and energy efficiency. The blueprint “will mark a point of no-return away from vertically integrated infrastructure,” said ABI’s Saadi.
Progress is being made on other fronts as well, albeit slowly:
- Vodafone last year switched on its first Open RAN-based 5G site in the U.K. as part of a plan to build out 2,500 such sites by 2027.
- Telefónica S.A. has said Open RAN will make up half of its new radio networks by 2025.
- Verizon Communications Inc. has deployed more than 8,000 virtualized cell sites toward a goal of having more than 20,000 in operation by the end of 2025. Although the sites aren’t based on Open RAN, virtualization is considered a key part of shifting to a more open set of protocols.
- Numerous wireless carriers in India, Thailand, Indonesia, Africa, the Middle East and South America are also conducting proofs-of-concept or are in early deployment.
In essence, though, most of the activity has been tire-kicking. “There’s no lead cheerleader for Open RAN,” said Gartner’s Fabre.
That said, technology trends may force the issue. Even as 5G is still creeping into the market, the next generation of cellular network technology, called 6G, is on the horizon with the promise of faster speeds, better coverage and improved reliability. Large-scale deployments are expected to begin by 2030.
While 5G set the stage for network virtualization, 6G is expected to make it mainstream. Virtualized networks are software-defined, allowing them to be managed and configured entirely in software. That not only dramatically reduces the need for field maintenance but gives operators unprecedented opportunities to precisely manage bandwidth and power consumption.
For example, carriers could temporarily allocate additional bandwidth to regions coping with a disaster or improve efficiency by turning towers on and off depending on the time of day or the day of the week, said Juniper’s Thattil.
“In 6G there’s just an assumption that the world will be virtualized,” said Wind River’s Cox. “The traditional vendors know that’s the case and so they’re having to adapt.” Rakuten has turned some heads with its claims that it manages its virtual network of 300,000 cells with a staff of just 250 people.
However, virtualizing off-the-shelf hardware that isn’t purpose-built for real-time processing “is daunting,” wrote Sachin Katti, senior vice president and general manager of the network and edge group at Intel in a recent blog post. “It is a very demanding piece of the network infrastructure and the cornerstone to ensuring your most important call is never dropped. It needs to deliver the high-speed and low-latency connectivity we all desire, but without ever failing.”
Several vendors, including Dell, Qualcomm, EdgeQ and AccelerComm Ltd., are working on or have delivered accelerator hardware to overcome the performance limitations of general-purpose CPUs.
In the long run, the RAN market won’t be a winner-take-all proposition. “It’s likely that Open RAN will coexist [with proprietary technology] for a long time,” said EdgeQ’s Ravuri. “The legacy will never go away and the tail will be very long, but I think in 10 years it will be at least 50% of the market.”
HPE’s Ram agreed that legacy RAN technology “is pretty vulnerable. [Carriers] are not going to turn on a switch and change things. But is it going to happen over time? Absolutely yes,” she said.
For enterprises, the tipping point could come when packaged services emerge that make deploying 5G and successor networks simple.
“There is a need to digitize the enterprise,” Saadi said. “This is something they are very keen about, but what they are not keen about is the disaggregation.” He sees private wireless networks evolving toward a subscription model, much like consumer wireless services today. “Selling equipment on a capex basis won’t work; it has to be as a service,” he said.
Once that and other obstacles get whittled away, deploying that private wireless network in a cornfield may become a whole lot more attractive.