A high-stakes Jenga game

Open RAN’s progress reflects a delicate balance between promises and practical hurdles, including slow adoption and single-vendor imperialism.

In an ideal world, it should have been like a fun and stimulating game of Lego blocks. But, as the practical world would dictate, it turned into that delicate and tenuous game of Jenga, where removing one important block would bring down the whole structure—collapsing—like a house of cards.

No wonder the buzz at the last Mobile World Congress was more about a one-vendor deal than any jaw-dropping innovation. The USD 14 billion contract between AT&T and Ericsson, reportedly the largest in the Swedish company’s history, was the talk of the town. Bruno Zerbib, Chief Technology Officer, Orange even dared to ask in an interview: “If there is a democracy with only one candidate, is it really a democracy?”

“Two major deals—Ericsson for AT&T and Ericsson with Samsung for Verizon—show that the industry is yet to embrace a truly multi-vendor Open RAN ecosystem.”- SYLWIA KECHICHE, Senior Director – Industry Analysis, Opensignal

A Black and White Rubik’s Cube

When Open Radio Access Network (RAN) was launched, it was based on the idea that combining components from various small suppliers could provide an alternative to large suppliers like Huawei, Ericsson, Nokia, and ZTE. However, as John Strand, CEO of Strand Consult, explains, the reality has turned out to be quite different.

“Open RAN was marketed with the promise that network components could be pieced together from different manufacturers to provide alternatives to end-to-end solutions from larger vendors. Numerous Open RAN ‘trials’ were announced by operators worldwide. However, 294 mobile operators in 109 countries and territories have launched 3GPP-compatible 5G services, either mobile or fixed wireless access. Overall, Open RAN equipment accounts for only about 1–2% of the total 5G equipment installed and used by customers today,” he points out.

Regarding the operators that have fully committed to Open RAN, such as Rakuten in Japan, 1&1 in Germany, and Dish in the US, they have struggled to attract and retain customers. Strand explains that vendors like Parallel Wireless and, more recently, Airspan have faced financial challenges, illustrating the reality of Open RAN.

“The problem with Open RAN is not the state of the standard but the operationalisation of the available Open RAN solutions.”- MANISH GANGEY, Executive President – Product Line Management, HFCL

“The concept of Open RAN was introduced to create a more open and competitive environment within the telecommunications industry,” echoes Manish Mangal, Chief Technology Officer, Telecom and Global Business Head, Network Services, Tech Mahindra.

Mangal affirms the dominance of one or two vendors: “Open RAN design aims to shift away from traditional RAN systems that are often constrained by single-vendor dependencies. Nevertheless, the transition to a genuinely open RAN is fraught with challenges. While Open RAN promotes using different vendors to stimulate technological advancement and cost efficiency, the reality is somewhat different due to the persistence of ‘single-vendor imperialism.’ This phenomenon is exemplified by various deals where operators opt for single-vendor solutions to mitigate integration complexities and ensure reliable performance.”

Ask Sylwia Kechiche, Senior Director – Industry Analysis, Opensignal, about this, and she predicts that a truly multi-vendor Open RAN ecosystem is still some way off. “The concept of Open RAN is based on the desire to reduce vendor lock-in and create competition for smaller players by separating hardware and software components.”

She cites data to illustrate this further: “According to Heavy Reading’s 2023 Open RAN Operator Survey, 31% of respondents plan to use a new single vendor that is Open RAN compliant, closer to the existing RAN operating model. For instance, in the US, two major Open RAN deals involved a single vendor (Ericsson for AT&T) or two vendors (Ericsson and Samsung for Verizon), indicating that the industry has yet to fully embrace the concept of a truly multi-vendor Open  RAN ecosystem.”

Ashwinder Sethi, Partner, Analysys Mason, avers, “Both in India and global markets, Open RAN is a nascent area due to concerns from multiple operators, despite broad statements and commitments in the US and Europe. There are medium-term and long-term ambitions. We only see a nascent ecosystem, except for Rakuten, which has worked substantially in Open RAN. It is hard to make definitive statements until someone deploys it properly and maintains it for two to three years.”

“The path to a fully open network is complex, exemplified by deals where operators choose single vendors to avoid integration issues and ensure reliability.”- MANISH MANGAL, CTO – Telecom & Global Business Head, Network Services, Tech Mahindra

“Just look at India, which has talked much about Open RAN. The classic suppliers, Ericsson and Nokia, have secured the big contracts,” Strand argues. “The AT&T–Ericsson deal is another great example.”

“In India, Airtel, Jio, and Vi have conducted some pilots, but they are small-scale deployments,” Sethi concurs. “Also, the cost of ownership is not very different from a conventional OEM.”

While operators such as Jio and Airtel have taken steps toward adopting Open RAN, progress has been slow, mainly due to the perception that it is still maturing, explains Priyanka Kulkarni, Manager – Telecom, Media, and Technology Sector, Aranca. “To push adoption, the Indian government has recently developed the US-India Open RAN Acceleration Roadmap, aimed at facilitating interoperability and widespread deployment of Open RAN products. This roadmap is also seen as an effort to encourage Indian telcos to adopt and deploy  the technology.”

Sethi adds that it is interesting how Nokia and Ericsson have entered this space. “They have realised they might face a much smaller market if they do not participate and support wide-scale adoption. There are also OEMs in this ecosystem—such as Mavenir—and traditional ones like Nokia and Ericsson, which are entering this space now.”

“Single-vendor solutions are expected to drive the lion’s share of Open RAN revenues, as many operators still depend on traditional vendor solutions.”- PRIYANKA KULKARNI, Manager – Telecom, Media, & Technology Sector, Aranca

As Kulkarni assesses, the current Open RAN landscape is witnessing a rise in vendor acceptance beyond the traditional giants. “Companies like Mavenir, NEC, Parallel Wireless, Fujitsu, Altiostar, and Airspan are increasingly gaining traction globally.” While this could signal a change in the previously traditional vendor-dominated market, single-vendor solutions are still expected to drive the lion’s share of Open RAN revenues, as many operators still depend on traditional vendor solutions.

“Single-vendor Open RAN solutions are likely to account for 15–20% of overall RAN revenues in 2028, whereas multi-vendor solutions are forecast to represent just 5–10% of the total market,” she points out.

Chips and Blocks

What prevents RAN from being Open? There are numerous factors to consider.

Let us delve into the internals first. Silicon’s significance has become pivotal as telecom infrastructure moves towards openness and disaggregation. Much relies on how easily hardware and software can be segregated and combined in new (non-proprietary) permutations and combinations. This is where the debate between general-purpose accelerators and inline accelerators gains momentum.

A lot hinges on how open, flexible, and programmable processors are. This makes the choice between custom-made chips and general-purpose silicon crucial and consequential for openness. Should one opt for custom-built silicon for Layer 1 and let the rest of the RAN work on general-purpose silicon? However, Layer 1 is the real spine, and if that gets hard-coded, how open is ‘open’?

Locking in silicon while opening up other RAN areas is a futile endeavour. However, reshuffling things in areas like Layer 1 and base-station silicon is no easy decision, given the availability and openness of processors. With Ericsson’s venture into purpose-built silicon and alliances between Nokia and Marvell or Vodafone and Intel, the dynamics of silicon for the Telco industry have witnessed many twists and turns.

Then, there is another factor: software. One cannot overlook the unique and entangled relationship between hardware and software, which can perpetuate the continuity of single-vendor RAN setups.

Kulkarni agrees. “The relationship between hardware and software in RAN does play a role in the dominance of single vendors. Traditional RAN deployments often result in operators being locked into using hardware and software from a single vendor, making it challenging to switch vendors or alter the network.”

“This vendor lock-in is a form of dominion that the close coupling of hardware and software can perpetuate. However, the shift towards Open RAN fosters a more open and collaborative environment. It allows product designers to select the hardware and software solutions that work best for them, leading to improved product performance, innovation, and cost-effectiveness,”  she explains.

“Different processor architectures used by various vendors can lead to compatibility issues and hinder seamless integration of hardware responsible for base-band data processing and performance,” Kulkarni adds. “Front-haul specifications are another critical aspect of the Open RAN architecture, wherein the new front-haul protocol optimised for mMIMO allows for better performance but continues to be dominated by traditional vendors. Standardisation of front-haul interfaces while ensuring bandwidth requirements between DU and RU is crucial for interoperability and multi-vendor participation in Open RAN.”

“Open RAN brings competitive pricing for equipment, but from what we see now, it will take five to six years for the ecosystem to evolve.”- ASHWINDER SETHI, Partner, Analysys Mason

Manish Gangey, Executive President – Product Line Management, HFCL offers a specific perspective here. “Traditional architectures always had the advantage of tweaking the functionality to suit their system design guidelines. When some of these nuances in design have to be communicated to a third party, it becomes complicated. Over time, these nuances in design requirements would become better known for the individual system components to be more independent. But if this argument was used against opening up the interfaces (in an open-source model), it was only to gain time against what is inevitable.”

Squares Against Circles

Let us now shift our focus to other components. The lack of a fluid interface between the base-band and radio sides means RAN merely pays lip service to the word “open”.

Apart from intent and imperialism, the on-ground integration and interoperability problems also throw a wrench into the Open RAN idea. This is particularly challenging due to the need for appropriate APIs, software, and algorithms that can power the ‘openness’ of RAN – facilitating the connection of different pieces of software and hardware across many network layers.

Mangal explains that one-two vendor decisions are often influenced by the current limitations in the availability of fully interoperable and mature technologies within the Open RAN ecosystem. “The Brownfield operators, for instance, might begin their Open RAN journey with a single vendor due to existing legacy systems but aim to transition to a multivendor setup in the future.”

It is crucial to fix interoperability for Open RAN, especially at the Distributed Unit (DU)–Central Unit (CU) level and between DU and Radio Unit (RU). Is that happening? Will it happen?

Kechiche agrees that the interoperability challenges between different hardware and software stacks slow its implementation and lead operators to choose a  single supplier.

The problem with Open RAN is not the state of the standard but the operationalisation of an Open RAN solution points out Gangey. “The goal of Open RAN is to have open, interoperable standard interfaces beyond what 3GPP prescribes. Some interfaces are more developed than others; for example, open front-haul is more evolved than an E1 or E2 interface.”

As he further explains, operators are accustomed to single-point troubleshooting scenarios when they encounter issues in the network. Now, they have to determine which part of the solution is causing the problem in the network. This requires both skills and knowledge of the internal workings of the solution as a whole. “Therefore, you see a deal like AT&T–Ericsson where they feel they get the best of both worlds.”

Here, Strand sheds light on how operators purchasing Open RAN equipment primarily buy from Ericsson, Nokia, and Samsung, not from the smaller players claiming Open RAN would radically change the market. “Operators prefer one-stop solutions, not multi-vendor solutions.”

Mangal delves deeper. “Interoperability in Open RAN, particularly at the DU-CU level and between DU and RU, presents notable challenges, primarily due to the nascent stage of this technology. Achieving true interoperability involves adhering to specifications and ensuring that different hardware and software components work seamlessly in diverse service provider environments.” Additionally, new specifications such as Split 7.2x Cat-B ULPI for Massive MIMO front-haul and the abstraction layer for processors reflect ongoing efforts to resolve compatibility challenges.

Kulkarni breaks this into three key components of gNodeB: the central unit (CU), distributed unit (DU), and radio unit (RU). “Integrating and managing the intricate technicalities of the key components to ensure seamless communication across these interfaces in a multi-vendor ecosystem while simultaneously adapting to changes for a tailored network configuration increases the complexity of the system.”

Gangey contends, rightly so, that interoperability works well when both sides are transparent. “The DU-CU splits are not as challenging as DU-RU splits because, in the former case, both components typically come from the same vendor, whereas in the latter case, they can come from two vendors. The front-haul specs have evolved and are much better today compared to when they started, but it is more about how you can make two functions talk to each other without extensive and expensive code reworking.”

Standardised open interfaces are key to building high-performing networks by enabling different vendors and components to work together efficiently. However, gaining consensus on the best ways to achieve it has proven difficult at times, Kechiche points out the thorn in the pillow.

“Among these interfaces, front-haul is one of the most critical in Open RAN as it connects the radio units to the network, making it central to the performance of the air interface and the overall RAN efficiency. Open RAN-dependent systems and components that can be procured from various vendors require a considerable amount of system integration expertise to ensure successful deployment,” she says.

According to her, the level of difficulty associated with integration varies depending on the size and internal structure of the operators and whether there is a need to integrate with existing systems. “Typically, larger operators with in-house engineering resources tend to fare better than smaller ones that may not have access to local system integration skills.”

Mangal suggests that when assessing the openness of RAN, it is crucial to consider several additional factors beyond the basic framework of Open RAN. “Firstly, the integration of RAN Intelligent Controllers, xApps, and rApps enhances the network’s intelligence and functionality. Additionally, the ability of Open RAN to handle and analyse the increased volume of data must be scrutinised, especially in terms of data management and privacy concerns,” he says, reminding how security and interoperability are crucial emphasis areas too.

Other Pieces of The Puzzle

There is also a flurry of new technologies, like virtualisation, Cloud RAN, and Massive MIMO—but are they helping or hampering the cause of openness  in RAN?

“These technologies enhance openness by enabling flexibility and resource sharing but also pose challenges related to interoperability, integration, and vendor dependencies,” remarks Kechiche. “For example, Massive MIMO requires close coordination between RU and DU and involves stringent  performance requirements.”

Kulkarni, too, points out that while these technologies enable the disaggregation and innovation within Open RAN, they also introduce complexity and pose interoperability challenges that need to be carefully managed.

Gangey zooms in on the ‘why’ of Open RAN, which is attempting two major changes to the Radio Network: openness through open interfaces and disaggregation of hardware from software.

“This creates a complexity that people outside of operators usually underestimate. In the technology world, new concepts bring complexity to deployment and operations before they get to simplification. Here, the virtualisation and cloud concepts are alien to the network world, especially radio access networks; therefore, it would take time for these to be understood, mastered, and then deployed at scale. So, in some sense, the introduction of too many changes together has hampered adoption,” he points out.

So? What Topples Next?

Open RAN fosters more competition in the market, and increased competition is good for operators and customers, Sethi paints the big picture. “It also brings competitive pricing for equipment. But from what we see now, it will take five to six years for the ecosystem to develop towards Open RAN.”

The Open RAN segment requires effective multipronged collaboration among various vendors, notes Nityesh Bhatt, Professor – Information Management Area, Institute of Management, Nirma University. “Openness is a continuum that takes time to mature and can never be perfect. Interoperability, as seen in many sectors including SIM cards and IoT, is a result of numerous initiatives from the industry (R&D, resources, intent), government (policies), regulators, research organisations, and think tanks across the globe.”

As Sethi predicts: “In the next two to three years, we are not going to see massive adoption of Open RAN, except only limited deployments here and there.”

Ask Strand, and he reiterates the long-held argument with a close focus on costs. “It is certain that Open RAN has not delivered what it promised when it comes to cost savings, more competition, and low prices. It will remain a commercial disaster as we had predicted several years ago.”

“It is certain that Open RAN has not delivered on promises of cost savings, more competition, and low prices, and it will continue to be a commercial disaster.”- JOHN STRAND, CEO, Strand Consult

In the view of Kechiche, Open RAN represents an important new methodology for MNOs, promising lower capex and cost efficiency. However, reliability and security are two aspects that cannot be overlooked.

“The GSMA Intelligence Network Transformation Survey 2023 points to solution reliability, integration with existing operations, and systems integration as the top three deployment concerns for operators, all of which cannot be ignored. Replacing manual processes with zero-touch operations is one of the key advantages of Open RAN,” she reiterates

Open RAN initiatives are gaining global momentum, with trials and deployments occurring across various continents, including Asian countries like Japan, India, and Thailand, and others in Africa, Europe, and the USA; Kulkarni maintains an optimistic stance. “Japan is one of the leading markets where operators have partnered with multiple vendors for Open RAN deployments. For example, Rakuten Mobile has partnered with over 20 vendors like Altiostar, Cisco, Nokia, Intel, Mavenir, Quanta Cloud Technology, Sercomm, Tech Mahindra, Allot, and Innoeye, while NTT has joined hands with more than four vendors,” she stresses.

Similarly, KDDI has partnered with two vendors and is exploring more options. In the US, Dish Network has partnered with over 12 vendors and Inland Cellular with over four vendors. There are also examples from European and African markets where operators are using multiple vendors for the deployment of  O-RAN networks.

Adopting Open RAN requires a mindset shift and is more suitable for new operators starting from scratch, Kechiche dissects. “Rakuten Mobile is currently the only commercially available nationwide network deployed using Open RAN technology, making it the poster child for this new paradigm. However, beyond Rakuten, Open RAN deployments have been slow so far, mostly limited to proof-of-concept and trials, very much following the natural rhythm of technology upgrades.”

Some governments, such as the UK, are strong proponents of Open RAN, and so are several European operator groups, such as Deutsche Telekom, Orange, Telecom Italia, Telefonica, and Vodafone. For example, Vodafone plans to use Open RAN across 30% of its European footprint by 2030, including Open Front-haul as a new addition to its requirements on the single RAN part. “We should see more Open RAN networks towards the second half of the decade, with Open RAN fully baked into 6G standards,” Kechiche adds.

As Mangal illustrates, Europe, Japan, and the USA have been pioneers in adopting this technology, driven by regulatory support and industry initiatives that promote a diversified vendor ecosystem. “These regions exemplify how Open RAN can successfully address the traditional constraints of mobile network expansion. These include high costs and vendor lock-in, which can be achieved by leveraging cloudification and advanced intelligence within the RAN. The momentum for Open RAN is now spreading to other regions, including the Middle East, indicating broader acceptance and implementation across various markets.”

Kulkarni observes that despite advancements in the 5G market (considerations of 6G) and the current state of deployment, achieving an Open RAN state is still a long way off.

In a February 2023 paper on this subject, Vodafone and DOCOMO acknowledged that the operators are responsible for improving the journey. The two companies recommended appropriate paths for vendors to take depending on which part of the Open RAN jigsaw they are developing. It was also underscored here that automation alone is not enough. It is vital that all parties make an effort to collaborate and leave behind any protectionism, the companies had emphasised.

To add to that, experts have also pointed out how the layered nature of most networks in this industry could lead to vendors being boxed in as per those layers—like management, CUs, DUs, and RUs. So, some layers could be more ready for multiple vendors, and some not. And some vendors may be more ready for turning ‘open’, and some not.

Interesting! Could Open RAN turn into a game of Woodoku next? In an ideal world, again. 

By Pratima Harigunani

pratimah@cybermedia.co.in