Ericsson today supports only one flavor of network app technology, and AT&T is solely reliant on it.
From a distance, it looked like a slick PR exercise, designed to show the criticality of Ericsson for a US market that lacks viable homegrown alternatives. But last month’s media tour of Ericsson’s Texan factory, where automated systems appear to outnumber people, also gave AT&T a soapbox to proselytize about the openness of its $14 billion radio access network (RAN) deal with the Swedish manufacturer.
Critics smell a rat. In future, service management and orchestration (SMO) as well as RAN software will come exclusively from Ericsson, which seems likely to provide most of AT&T’s radio units (RUs), too (Fujitsu, an Ericsson partner, is the only other named RU vendor so far). Any real chance for supplier diversity, then, is in the RAN Intelligent Controller (RIC) and the apps it can support. Telco technology executives gush about these much as smartphone addicts enthuse about their Android and iPhone apps.
But take a few calming breaths before you get carried away about diversity. For starters, AT&T is not combining Ericsson’s RAN with a third-party RIC, a mash-up that open RAN’s interoperability should allow. Instead, it is using Ericsson for both. That’s largely because it has opted for a non-real-time RIC that is housed within Ericsson’s SMO technology, branded the Intelligent Automation Platform (IAP).
IAP isn’t restricted to Ericsson’s RAN, the Swedish company insists. In true open RAN style, it should function with any other vendor’s RAN technology, whether purpose-built or virtualized. But concerns about standardized specifications have been publicly aired. Most notably, in a white paper published in late 2023, Vodafone and NTT Docomo complained that interfaces to other RAN elements, including the O1 link between the SMO platform and RAN functions, were still not ready. “SMO interface specifications need to be finalized as soon as possible to avoid prolonging vendor lock-in and delaying the widespread adoption of open RAN,” said the authors.
What’s unclear is whether there has been much progress in recent months. But in February, at Mobile World Congress, Nokia’s Tommi Uitto, who runs the Finnish company’s mobile networks business group, said any vendor providing both the SMO platform and the RAN software would be able to create “sticky functionality” between them. In other words, replacing the SMO platform or the RAN software might be harder if both come from the same vendor.
Regardless, AT&T is not even attempting to unite an SMO or RIC platform from one company with RAN software from another, and there is no sign it plans any such combo in future. All this makes it even more heavily reliant on Ericsson. What’s more, the RIC has increasingly been likened to an app store for the network. Yes, Ericsson’s will host various third-party apps, as AT&T proclaimed in Dallas last month, just as Apple’s does for consumers. Will it also, then, extract Apple-like fees from app developers?
rApps or nothing
The RIC flavor won’t be to everyone’s taste, either. Just over a year ago, AT&T claimed to be “excited” about RIC work it was then doing with Nokia. At the time, the Finnish company said it was the only big vendor offering support for E2, an interface that opens network functions to a near-real-time RIC. This differs from the non-real-time RIC cocooned in Ericsson’s IAP by allowing updates to be done in milliseconds rather than seconds.
Does that matter? The industry seems to think so, using the xApps label to distinguish this category from the seemingly more pedestrian rApps of the non-real-time RIC. But AT&T’s interest in xApps seems to have waned since it decided to replace Nokia across one third of its network with Ericsson, which already served the other two thirds. Unlike Nokia, Ericsson has made no accommodation with xApps and does not have a near-real-time RIC of its own to show off. This leaves AT&T with rApps and nothing else.
There could be valid technical reasons for Ericsson’s current aversion to xApps. Specifications have not advanced as much in the near-real-time RIC as they have in the rApps world, according to analysts and industry experts. Ericsson is not the only big kit vendor that does not support xApps. Samsung, the third-biggest RAN vendor outside China, after Ericsson and Nokia, has also based its in-house strategy exclusively on rApps until now.
Ericsson explains its approach by saying xApps still present “interoperability challenges.” It is “following the evolution of standards,” it told Light Reading, but sounds unconvinced of the need for a near-real-time RIC. “The functionality of xApps is provided by the RAN itself,” said a spokesperson by email. “At this point, we don’t see benefits of xApps on top of what our RAN is doing.”
But suspicious minds think Ericsson is worried about exposing network nodes to third-party innovation. Perhaps the most obvious threat comes from Cohere Technologies, arguably the most prominent xApp developer. Its Universal Spectrum Multiplier, available as an xApp, is designed to boost network capacity on radios that have already been deployed, and its performance has been lauded by Deutsche Telekom, Vodafone and other telcos after lab and field trials. Cohere feasibly makes xApps a problem for a vendor that would naturally prefer to sell more expensive radios than see older ones supercharged by some third-party software.
None of this totally rules out an eventual AT&T shift to xApps. But the deployment of any third-party near-real-time RIC would mean bolting on an additional platform and linking it to Ericsson’s systems and functions, presumably through interfaces such as E2. Ericsson could always pivot, surprising everyone with a move to support xApps through its own or a third-party platform. But that would pose risks to its business.
One analyst thinks Ericsson in open RAN is like the kid in the school bus who quietly sits at the back each semester. Nobody ever speaks with him until one day he is driving the bus ā and steering it in a completely different direction. Where AT&T ends up is anyone’s guess.
Original article can be seen at: