Robotic AI fine tunes on the fly
Nokia has broken new ground in the radio access network (RAN) field, after becoming the first major radio access equipment maker to run applications under the auspices of the instant RAN Intelligent Controller (RIC) and E2 interface. This accord between machines means the RAN can run applications with native Open RAN compliance. This in turn means operators can use AI to micro-manage a network to within a millisecond of its best possible performance.
The vendor’s engineers have toiled tirelessly with their peers from US telco AT&T to create a working model. In March they triumphed with a trial run that validated the promise of the ‘near real-time RIC’ and ‘xApp approach’ for advanced 5G use cases. “This adds a new intelligence layer to the radio network,” said Mark Atkinson, Head of Radio Access Networks at Nokia.
The breakthrough was made using Nokia’s commercial near instant intelligent controller of RANs, AKA RIC, running on Nokia AirScale base stations on AT&T’s network. In this trial, ‘near real-time xApps’ used E2SM Policy Services to ‘dynamically’ get the best possible outcome out of all the many variable that comprise the radio access network, at any particular microsecond. Near real-time RIC is discriminating and can finesse the service offered to specific User Groups, Frequencies or any other of the quality of service (QoS) class Identifiers inherent in 5G networks. In the interests of continuity, Nokia’s near real-time RIC system and xApps use existing interfaces to fine tune RAN the use cases according to the operator’s network choices.
As mobile connections proliferate operators must engineer more efficient radio network performance, said Nokia, and near real-time RIC gives ‘xApps’ the option of using artificial intelligence and machine learning algorithms that constantly push RAN performance. Nokia’s Advanced Traffic Steering xApp, for example, spreads mobile traffic across different frequency layers, freeing up bandwidth for mobile users. Meanwhile the Anomaly Detection xApp uses machine learning to spot irregular behaviour patterns in the RAN.
The trial proved that the E2 interface can update RAN policies many times quicker than through the old OAM interfaces, according to Robert Soni, an AT&T VP for RAN Technology. This “unlocks new methods for RAN optimisation,” said Soni.
Near real-time RIC has the potential to become a key enabler for RAN programmability, according to Nokia’s Mark Atkinson, “Open collaboration drives innovation across the telecom industry and harnesses the true power of 5G.”
Read more about Nokia’s Service Enablement Platform