By Robert Froehlich and John English
Open Radio Access Network (O-RAN) disaggregates the RAN into cloud-based service functional systems in which the usually integrated and proprietary base station is broken into three service components: radio unit (RU), distributed unit (DU), and central unit (CU).
There are several good reasons for O-RAN. First of all, O-RAN promises a best-of-breed approach that lets mobile operators pick the radio from one vendor, the distribution unit from another, and so on. In that way, the mobile operator’s RAN is not subject to the costs, product limitations, and delivery timeline of one vendor for the entire base station.
Meanwhile, multi-access edge computing further drives the need for a disaggregated RAN, with a range of enterprise use cases in which both the RAN and the core functions may be divided between the enterprise premise, the access network, and the core network as required by the 5G service application. And finally, for the operation of a cloud RAN, artificial intelligence (AI)-enabled RAN services and O-RAN interfaces for policy will be essential.
All of this sounds great. But of course, these benefits come with new challenges.
Looking at the transport layer through the CU, DU, and RU, early adopters are reporting IP transport issues for basic interconnection between the various RAN functions that have been “broken up.” At the same time, carriers are reporting security concerns associated with cloud management of IP transport.
With the O-RAN architecture, operators are expecting the CU to support as many as hundreds of DUs. Accordingly, they must manage CU/DU scaling and DU performance along with inter/intra-cell handover performance and vendor interoperation, all in a cloud environment.
For the control plane, carriers must look at CU performance, inter/intra-CU handover issues, and vendor interoperability. For the user plane, the focus must be on performance and on ensuring QoS/QoE to achieve expected levels of 5G services.
The RAN intelligence controller (RIC) poses further challenges for RAN operations. At present, the RIC has only high-level insights within a (radio resource) connection and does not have the ability to understand the connections between the same subscriber or other user equipments (UE). With no IMSI/IMEI correlation across connections (for both specific and multiple UEs), there is increased risk for false positive alarms/actions. For example, if a handset type causes many problems in a few cells, the RIC may not be able to understand and isolate that specific device issue and instead may falsely identify a systemic network issue.
Finally, the vendor RU/DU/CU mix and match poses potential interoperation issues as vendors deviate from standards to differentiate themselves. How do the various integrated components work together in live network operation?
Integration, Deployment, and Operational Risk
The 5G race is on, with much commercial pressure to launch. But with an integrated cloud RAN, mobile operators should anticipate unexpected issues that will increase time to market, as well as unexpected network performance degradation. Having cloud-optimized instrumentation to gain visibility will give operations teams the ability to proactively monitor and optimize 5G New Radio (NR) performance. With an explosion of cells (five to ten times more than with 4G) expected in O-RAN, propagation modeling for 5G NR design and operational automation will be needed to reduce costs. Smart data is needed to verify orchestration, because typical AI input/learning often leads to output errors with no way to understand the decision path.
Complexity bests summarizes the challenge of the new multivendor cloud-based implementation of 5G O-RAN. Having complete visibility for operation, orchestration, analytics, and security throughout the 5G lifecycle is critical to 5G success and will enable mobile operators to realize the vision of 5G for a lower-cost, agile, multivendor, cloud-based RAN.
Learn more about NETSCOUT's O-RAN Solutions