5G is a key driving force behind the migration to open radio access networks (O-RAN). Network densification will increase exponentially with 5G and mobile operators need to adopt standards-based networks to improve resource utilization. In addition, the existing network infrastructure and traditional fronthaul technologies are just not able to address the increased demand and bandwidth coming with 5G. Mobile network operators need O-RAN to evolve to 5G.
In parallel, innovative technologies such as virtualization and software-defined networking (SDN) have matured. These technologies can now be integrated into radio access networks. With O-RAN, mobile network operators can maximize their network performance and optimize their capital and operational expenditures (CAPEX/OPEX). Rakuten Mobile, for example, expects to save 30% on CAPEX and 28% on OPEX by migrating to O-RAN (Reference 1).
O-RAN offers significant potential to mobile network operators. An open network fosters innovation by letting in new entrants, and these networks are also standards-based, enabling interoperability between equipment from different vendors.
The O-RAN paradigm
While providing greater flexibility in how networks are built, O-RAN also represents a paradigm shift for mobile network operators. Until now, a base station was a single entity coming from a single network vendor, so it was tested as a unit. With O-RAN, the base station consists of different components coming from different vendors; even subcomponents may come from different suppliers.
The main components of an O-RAN include:
- The central unit (O-CU), which handles the packet data convergence protocol (PDCP) layer of the protocol between the network and the user equipment.
- The distributed unit (O-DU), which handles the medium access layer (MAC), radio link layer (RLC), and the high physical (H-PHY) layer.
- The radio unit (O-RU), which performs the fast Fourier transforms (FFT), some beamforming, and precoding.
- The RAN intelligent controllers (RIC), which include the non-real-time RIC that gathers information from the O-DUs, O-CUs, and O-RUs, performs analytics, and uses machine learning (ML) and artificial intelligence (AI) to generate suggestions for these network elements to optimize the RAN; and the near real-time RIC that pushes the commands to the network elements.
Figure 1 An O-RAN includes various components.
Mobile network operators need to put these elements together and ensure that the components they source from different vendors work with each other seamlessly. Even though the specifications in place are detailed, there is room for interpretation, creating challenges.
O-DU testing challenges
While interoperability testing is necessary for all O-RAN components, DUs present significant challenges from this perspective. To start, management plane (M-plane) implementation can cause issues because the DU expects certain parameters when the M-plane is implemented and stops working if these conditions are not met. This can be a significant challenge because the YANG models that help manage the radio units feature more than 6,000 parameters, with less than 3% of them mandatory, and network vendors also implement custom protocols.
There is also a lot of flexibility in the O-RAN specification for the bring-up sequence. Some DUs skip the get-schema process or stop working if the RU behavior differs from their expectations.
DU units can also stop operating if certain protocol options are missing. For example, some DUs stop operating if the dynamic host configuration protocol (DHCP) options are not exactly what they expect.
The O-RAN specifications include many implementations for the control plane (C-plane) and the user place (U-plane), and the mapping of these planes will change depending on the type of RU. DUs are designed to work with a specific RU type, beamforming model, and compression rate. A DU typically supports either “category A” O-RUs, which are non-precoding radio units or “category B” O-RUs, which are pre-coding units that support modulation compression. They also typically support only one beamforming model — pre-defined, weight-based, attribute, or channel — and each model has specific methods and options.
All these options are great for innovation but the variety of protocol implementations creates a significant challenge from a testing standpoint. In addition, validating fronthaul timing is critical. The O-RU needs to be tightly synchronized to the DU, for features like time division duplexing (TDD), multiple-input/multiple-output (MIMO), and multi-RU carrier aggregation.
Finally, multi-user MIMO (MU-MIMO) requires close cooperation with the O-DU vendor. The O-RAN Alliance has standardized communications across the O-RAN interface, but the use of sounding reference signal (SRS) and beam weights remains vendor-specific.
O-RAN testing solutions and use cases
To overcome these challenges, you will need a test solution that provides broad capabilities to cover the RUs from different vendors. M-plane flexibility requires the simulation of multiple behaviors from different RUs. Your test solution also needs to support both category A and B O-RUs and synchronization plane (S-plane) features, as per the O-RAN.WG4.CUS.0-v05.00 specification, so you can validate fronthaul synchronization. Aligning with specific implementations requires support for beamforming and MU-MIMO adaptation.
Figure 2 shows a possible configuration for 5G non-standalone (NSA) O-DU testing using a real evolved packet core (EPC) network.
Figure 2 An RU simulator and a traffic generator enable 5G NSA O-DU testing.
In this configuration, the RU simulator connects to the DU via the fronthaul eCPRI interface and uses the LTE anchor over the X2 interface to connect to the CU. The simulator uses the S1-AP interface to connect to the core network. A traffic generator provides the load on the network.
For 5G standalone (SA) testing, the configuration can be much simpler as the RU simulator does not need to emulate the eNB (Figure 3). The RU simulator just connects to the DU using the eCPRI interface.
Figure 3 An RU simulator and traffic generator provide a simpler configuration for 5G SA O-DU testing.
However, complete O-DU testing requires wrap-around testing. Figure 4 shows an example configuration for wrap-around O-DU testing with a simulated EPC in a 5G NSA context.
Figure 4 An RU simulator connects to a core network simulator to enable O-DU wrap-around testing in a 5G NSA environment.
The RU simulator connects to the DU via the eCPRI interface and to the CU via the LTE anchor over the X2 interface, but uses the S1-AP interface to connect to a simulator that emulates the core network.
Figure 5 shows the setup for 5G SA with a simulated 5G core network. As in Figure 2, the RU simulator connects to the DU using the eCPRI interface, but a simulator simulates the 5G core network.
Figure 5 An RU simulator and core network simulator enables O-DU testing in a 5G SA environment.
Overcoming O-DU interoperability testing challenges
The O-RAN movement is gaining momentum. Companies building equipment based on the O-RAN standards are leading to rapid improvements in the specification. The greater flexibility brought by O-RAN also represents a paradigm shift for mobile network operators that now need to put the various network elements together. Ensuring that the components sourced from different vendors work together seamlessly has become a top priority. Interoperability testing can be challenging, especially for the DU due to high flexibility in protocol implementations and the need for tight synchronization with the RU. Test solutions that provide broad capabilities to cover the RUs from different vendors are critical to succeed at O-DU interoperability testing.
For more information on O-RAN challenges and solutions, visit Keysight’s O-RAN page.
Jessy Cavazos is part of Keysight’s Industry Solutions Marketing team.
- Mobile World Live webinar “Why Open RAN Makes Sense: Facts from Real-World Experience”, March 11, 2021.