• AT&T scientist Matti Hiltunen is focused on RAN Intelligent Controller R&D
• He believes the RIC offers significant advances in terms of radio access network management and control
• A number of RIC app use cases have been identified that will enhance RAN deployments
• But AT&T is still looking for the “killer use case”

Open RAN is transforming the radio access network (RAN) and will have “a huge impact on how RAN intelligence can be implemented,” according to a senior member of AT&T’s technical staff.

Speaking during the O-RAN Alliance’s recent webcast, Matti Hiltunen, lead inventive scientist at AT&T Labs Research, said he believes the deployment of Open RAN architectures with RAN Intelligent Controller (RIC) elements will significantly advance the way network operators can control their networks to become more reactive, more efficient and improve the customer experience.   

“Currently, we are limited to implementing RAN intelligence either in the RAN itself or in central CSON [centralised self-organising network] platforms,” he noted. “While the CSON platforms are good for machine learning (ML)-type algorithms, they are limited by the latency of data collection and control. The RAN itself is resource limited for ML-type algorithms and, of course, from an operator's point of view, it's a closed system,” explained the AT&T man.  

An Open RAN architecture introduces two new elements – the near real-time RIC, and the SMO (service management and orchestration) software plus non real-time RIC. These RIC platforms enable “RAN intelligence to be implemented as apps”: xApps, in the case of the near real-time RIC, and rApps, in the case of the non real-time RIC. The introduction of the RIC systems requires the development of new interfaces, the A1 and E2, which “allow apps to have near real-time control of the RAN”, something that operators don’t currently have. (The A1 interface connects the two types of RIC, and the E2 interface connects the near real-time RIC and the RAN elements – the centralised unit (CU), distributed unit (DU) and radio head.)

Hiltunen believes the near real-time RIC ecosystem is in “good shape,” though the development of the application programming interface (API) that enables xApps to run on the near real-time RIC has only just started and “there is a lot of work to do in that space”. 

That confirms what TelecomTV has been hearing from many other experts in the industry: The good news is that it is still very early days for the development and implementation of RIC systems, which aren’t expected to be introduced into commercial networks until late 2023 at the earliest.

That doesn’t mean xApps and RIC platforms haven’t already been developed. Hiltunen noted that the xApp ecosystem has started to take shape. “There are a number of companies who realise they have proprietary technology or ideas that they can implement as xApps that can run on other companies’ RIC platforms... currently, the ecosystem is forming around each RIC platform implementation. But with the standardisation of the RIC API, hopefully the xApps will be portable soon,” he noted.

For AT&T, there are a number of key use cases where xApps can provide near real-time control of the RAN in a very positive way, and the operator has been proposing and supporting use cases in the O-RAN Alliance working groups and trialing various applications, either in open-source groups or internally in its own labs.

The use cases fall into multiple categories. “Some improve RAN efficiency, for example, improving throughput or spectral efficiency – traffic steering and MIMO are examples of such use cases,” explained Hiltunen. For improving the user experience, xApps related to network slicing and improving video quality are of interest to AT&T, while operational costs are “important for all operators, so the energy-saving use case that is currently being discussed in [O-RAN Alliance] working group 1 is getting a lot of attention.” 

And then, of course, there is security. Hiltunen noted particular work on “anomaly detection” and highlighted the “signalling storm prediction use case” that is being worked on by Alliance members.

Despite all of these important use case avenues, Hiltunen admitted: “We're still looking for the killer use case that would help justify the cost of the new network element. Naturally, after the RIC has been deployed, a number of use cases will be running on the same platform, but the initial use case has to be a strong one.”

And there’s a lot of work to do and a long way to go before RIC platforms are emerging from operator and vendor labs. “There are still many aspects that need to be proven by the RIC vendors and more work that needs to be done in the O-RAN community,” said the AT&T man. Scalability, how a RIC is managed, how xApps from multiple developers can be integrated, the security implications of deploying a new platform and running third-party apps, and the portability of xApps between different RIC platforms are all areas that Hiltunen identified as still requiring a great deal of work.

In addition, he noted, the RAN equipment vendors have yet to introduce support for the E2 interface that is vital for open, standardised communications between the near real-time RIC systems and the CU and DU elements.

Despite the sobering realisation of just how much still needs to be done, Hiltunen is adamant that the Open RAN era will be a step forward for mobile network operators.

“The Open RAN architecture is really transforming the RAN,” he said. “It's providing an open ecosystem for third-party xApp and rApp developers and this ensures fast innovation and RAN customisation” for particular deployments. For example, “imagine customising a private network that is in a factory setting, or optimising the network for special events such as the Super Bowl,” he enthused.  

O-RAN standards are “maturing and RIC and SMO platforms are emerging”, he concluded, adding that “a number of operators have done RIC trials that are showing promising results for many use cases.”

- Ray Le Maistre, Editorial Director, TelecomTV