In earlier blogs, I’ve discussed the use cases that 5G is likely to support across industries in the long run. I’ve particularly focused on the very diverse requirements associated with use cases for enhanced mobile broadband (eMBB), ultra-reliable low-latency communications (URLLC), and massive machine-type communications (mMTC). All these use cases share edge deployments as the common pattern.
RAN – from Data Center to the Edge
Prior to 5G, full-stack services for mobile customers were made possible by a mix of custom software and hardware components that were hosted in data centers (DCs). In the past, DCs contained security platforms like firewalls along with packet core components, administration duties, authentication, and authorization tools. Network designs created up until recently managed these operations from a small number of central locations, known as the main DC or the central DC. However, the deployment of these data centers in the foreseeable future is a result of the changing 5G service requirements as well as the decomposition of the mobile core and radio access network (RAN). Since the network transport is more closely entwined with the DC infrastructure in this distributed DC paradigm, DC placement and integration must be taken into account while designing any transport network especially RAN. The key message here is that Edge datacenters or cell sites shown at the bottom of the below picture are now the distributed DC for RAN deployments.
This figure summarizes the O-RAN-defined planes of operation (control (C-plane), management (M-plane), user (U-plane), and synchronization). The decomposed RAN (RU, DU, and CU) fragments are dispersed across the xHaul (fronthaul, midhaul, and backhaul) networks, which are not depicted in the picture. The bandwidth, latency, and device capabilities of each of these xHaul networks are constrained in different ways, which affect the physical topology and network design. The RAN and 5GC deployment architecture chosen will determine the position, size, and even the existence of these DCs.
The graphic also illustrates the 5GC’s control and user plane separation as well as the offloading of user traffic to be handled by MEC applications or distributed peering closer to the mobile subscriber. The transport network connects and combines many planes of operations spanning the far-edge, edge, and local data centers that house the mobile services.
The design of the transport network is thus strongly impacted by far-edge and edge DCs.
RAN Requirements
So in the context of the above what are the key features included in 5G NR and what challenges do they present to cloud designers?
- Increased number of cellsites due to per unit smaller coverage areas. Thus we will have denser radio deployments approaching thousands per metro
- A need to support both high bandwidth and low latency applications at quicker speeds
- A continuing architectural change toward what is usually referred to as a vRAN architecture has been caused by the decomposition and virtualization of the previously co-located RAN components.
- The design of the transport network itself at far-edge and edge DCs.
With the arrival of 5G, Radio Access Networks (RAN) deployments are going to be based on a combination of (a little bit of) virtualization and (a lot of) cloud-native technology such as containers. These broadly promise to address the wide range of needs for resilient, and intelligent mobile networks but any telco workload that can benefit from running in the cloud will eventually do so. One needs to bear in mind that this will be a gradual evolution as every CSP has a different network evolution journey, and a different ideal transformation speed. Net net, whether cloud (region) RAN or edge RAN – both offer the tools required to make new 5G use cases and applications possible.
The next blog will examine how we can combine the two architectures (Cloud and Edge) effectively to make the most of each deployment.
Featured Image by Erich Westendarp from Pixabay