During the RCR Wireless News‘ Intelligent RAN Forum 2026, Viavi marketing manager Owen O’Donnell and Battelle vice president of product development Mark Reudink examined the evolution of open and intelligent RAN
In sum – what to know
Openness expands innovation – Open RAN is bringing more vendors and solutions into the ecosystem, while network intelligence enables new use cases.
Integration accelerates – Battelle’s Mark Reudink said integration timelines for active antenna units have fallen from months to weeks as vendor software and testing platforms mature.
AI needs validation – Viavi’s Owen O’Donnell said testing is critical as AI applications influence network decisions and operators benchmark competing solutions.
Open RAN and network intelligence are increasingly developing together, expanding the number of vendors, applications and use cases across the RAN ecosystem. But as intelligent applications become more prominent, interoperability, integration, and testing remain important challenges.
The issues were discussed during the RCR Wireless News‘ Intelligent RAN Forum 2026, where Viavi Solutions marketing manager Owen O’Donnell and Battelle vice president of product development Mark Reudink examined the evolution of open and intelligent RAN.
For O’Donnell, openness and network intelligence can provide operators with complementary benefits, including greater vendor participation and progress toward more autonomous networks. “I think, in a way, openness and network intelligence is nearly a perfect storm for the operators because the openness is bringing in new players, it’s increasing the playing field, it’s reducing the cost, while network intelligence is allowing operators to head towards the goal of fully autonomous networks,” O’Donnell said.
He said potential benefits include reduced capex and opex, faster time to market, more streamlined network operations, and increased competitiveness through innovation.
Reudink said the Open RAN ecosystem has also matured from an integration perspective. He said Battelle’s first integration of its active antenna radio unit with a distributed unit took months, but that timeline has subsequently fallen to weeks.
Reudink also pointed to applications enabled by the broader ecosystem, including network steering and location solutions for factory floors where GPS may not be available.
O’Donnell said the RIC has moved from being a specification on paper toward practical testing in production labs, while a growing range of use cases includes network-slice assurance, massive MIMO, signaling-storm detection, SMO scale testing, drone-swarm detection, and application-collision testing.
O’Donnell also cited a Viavi and NTT Docomo demonstration involving AI-driven beamforming control that could improve system throughput by up to 20% by reducing control overhead.
Testing remains important as Open RAN deployments become more complex, O’Donnell said. “The challenges are interoperability, integration, compliance and standards, security, increased attack surface, and regular software updates,” he said. “So they’re all the primary challenges that come with Open RAN.”
He said trust in AI applications is another major issue, particularly when applications are influencing network decisions without a human in the loop. AI applications are only as good as the data used to train them, while operator network data can be proprietary and confidential.
O’Donnell said synthetic data can provide an alternative. Viavi’s AI RAN Scenario Generator can create RAN scenarios, run iterations to train AI models and then validate their performance against scenarios.
Reudink also emphasized the importance of common data approaches across vendors. “The data that comes in is fundamental to be able to have accurate AI models, to be able to really implement changes in that network, be it anything from smarter beam management to network steering,” he said.
The speakers also discussed the evolution of the AI-RAN Alliance and the move toward greater collaboration between vendors, operators, and other industry participants. Reudink said standardized metrics and data are important as multiple vendors and systems become involved.
The discussion also highlighted the role of multi-vendor testing environments. O’Donnell said operators can benchmark competing AI applications in their own labs using real or tailored network scenarios, including evaluating energy savings alongside quality-of-experience impacts.
