By Joe Mocerino, Fujitsu Network Communications
Planning and deploying a mobile network to support a myriad of 5G applications will be no easy feat, considering the complexities of these new architectures and the interdependencies between the RAN and transport network.
RAN transport rates for 5G will be over 15 times greater than those available in 4G LTE and its variants.
However, using the same operation of the 4G RAN for 5G would yield transport rates for optics and platforms price prohibitive. To mitigate this situation, the 3GPP standardized a new RAN model splitting the processing functionality of the 5G BBU into several blocks, thereby reducing the transport rate requirements.
The key building blocks of the Next Generation RAN (NG-RAN) architecture are the centralized unit (CU), distributed unit (DU) and remote radio unit (RU). Front haul transport between the RU and DU will use the more efficient eCPRI protocol which provides higher performance at a lower cost per bit than CPRI used for 4G services.
The IEEE has standardized the latency budgets for the new 5G RAN. These budgets are similar to the original 4G front haul and back haul segments with the exception of the new ultra-reliable-low-latency connectivity (uRLLC) use cases. In the front haul, these result in a 50 microsecond latency budget.
Back haul remains at 10 milliseconds and front haul at 100 microseconds for all but uRLLC applications.
The new area of transport is the “midhaul” or next generation fronthaul II (NGFH II) section, which will vary from one to three milliseconds in latency budget as per the IEEE 1914.3.
With 5G services, a new form of RAN topology is emerging. The predominate topology in the RAN today is the distributed RAN. The distributed RAN consists of all the 4G elements- remote radio head (RRH) and base band unit (BBU) at the cell site. This topology has the lowest latency. Next is the centralized RAN where the BBU is centralized at a location within 20 kilometers of the cell site. The centralized RAN configuration enables the BBUs at the central location to pool resources to address the demands of the cell sites. This eliminates the risk of over or under engineering the individual cell site with a specific capacity of BBU. Cell site aggregation also enables two or more cell sites to address demands of an individual mobile user.
The “virtualized” RAN is the new model for 5G. The processing elements of the RAN, i.e. the DU and CU- will ultimately be virtualized because vertical network slicing will initiate in the DU. This is the most flexible topology as it can be dynamically repurposed. The Next Generation Mobile Networks (NGMN) consortium of service providers has developed several RAN topologies. These models vary from a distributed RAN with site cost and complexity balanced with less demanding transport to a centralized RAN, which provides a coordination gain and yields a high-performance transport layer.
This flexibility is accomplished by splitting the functionality of the 5G RAN elements to deliver the performance requirements needed for the upcoming 5G use cases. The virtual RAN and transport topology will work very closely together. Service providers will be able to develop a single infrastructure to address the upcoming use cases and multi-tenant operation without having to dedicate assets to one topology type or having multiple network element overlays.