Cloud RAN and the use of G.fast in 4G and 5G networks
BT researchers, in collaboration with semiconductor manufacturer Cavium Inc, have demonstrated Cloud RAN (C-RAN) over G.fast - an experiment that is believed to be a world first. But what is C-RAN and why is the result significant?
Cloud RAN, or cloud radio access network, is the latest step in the evolution of radio access architecture for mobile networks. This article gives a high-level overview of this evolution and explains why the C-RAN/G.fast result could be important for the rollout of 4G and 5G networks.
Radio Access Network (RAN) evolution
Early radio access networks were based on 'all-in-one' cabinets at mobile network base-station sites (fig.1). These cabinets contained two major elements (in addition to associated power systems and backhaul circuitry): a Baseband Unit (BBU) to process user and control data, and a Radio Unit (RU) to generate the radio frequency (RF) signals to be transmitted via the cell site's antennas (typically located on a cellular mast or roof top). The radio unit would be linked to the antennas via coaxial cable. There are two main problems with this design: signal loss in the coax cable and potential radio interference from nearby basestations.
For large or 'macro' basestation sites, a fibre backhaul link typically connects the baseband unit to the mobile operator's core network.
Fig.1 Traditional 'all-in-one' basestation cabinet
Distributed RAN (D-RAN) - The first evolutionary step was to separate the base-station equipment (fig.2). The radio unit was moved closer to the antennas to become a 'remote' radio head (RRH). This solved the coax signal loss problem. In this configuration a fibre optic cable is used to connect the RRH to the BBU.
Fig.2 Distributed RAN
Centralised RAN (C-RAN) - Having separated the BBU from the RRH, the next logical step was to move the BBU further back towards the core network, this could be several kilometres away from the cell site. This configuration enables multiple BBUs to be grouped together in a centralised location, sometimes referred to as a 'baseband hotel'. With the BBUs co-located in this way, it became easier for the radio transmissions of different basestations to be co-ordinated – reducing potential interference and maximising spectral efficiency. This architecture is also known as the 'fronthaul model' – fronthaul being the link between the BBU and the RRH (fig.3).
Fig.3 Centralised RAN
Cloud RAN (also C-RAN) – The final step is to use Network Functions Virtualisation, or NFV, to 'virtualise' the BBU functionality using generic computing hardware to give 'cloud RAN' (fig.4). Some people also refer to this as Virtualised RAN (V-RAN).
Fig.4 Cloud RAN
Benefits of C-RAN
C-RAN lets operators improve network performance and reduce the costs of deployment and ongoing operations.
Consolidation of BBU functions at data centres nearer to the mobile core leads to:
- reduced basestation footprint
- simpler, cheaper and quicker network expansion - the mobile operator only has to install RRHs at new cell sites and connect them to the centralised BBUs
- fewer baseband units serving multiple RRHs
- reduced interference through better coordination of basestations
- reduced energy consumption (through the sharing of power-consuming equipment such as air conditioning)
Virtualisation (cloud RAN) gives further benefits:
- the BBU functionality is hosted on generic computing hardware allowing data processing resources to be 'pooled' (shared across multiple cell sites)
- resources can be allocated in response to fluctuations in bandwidth demand, ensuring reliable operation of the network through load balancing
- scalability - as data usage grows and traffic patterns evolve, increased processing capacity can be easily provided by via hardware upgrades in the data centre
The potential role of G.fast in the fronthaul
To date, C-RAN has required a dedicated fibre fronthaul link to connect the transmitters (RRHs) at the cell sites to the signal-processing functionality deeper in the network. This will continue to be the case for high-capacity macro cell sites and it can involve complex and costly engineering work if no fibres are present in the ground to carry the signal.
But as mobile operators continue to build out 4G networks and look ahead to 5G, the concept of 'ultra-dense networks' comes into play. These networks consist of deployments of large numbers of small cells, located for example on lamp posts (fig.5). These small cells won't require the capacity of a large macro cell but to be commercially viable they'll probably need a more economic fronthaul connection.
Fig.5 Cloud RAN with G.fast fronthaul
If the signals to/from small cells could be carried over copper lines it would remove the need for dedicated fibre connections. A cloud RAN service delivered over G.fast for example would significantly lower the cost of deployment for mobile operators building out 4G and 5G networks. And that is why the BT/Cavium result is important. The trials demonstrated that G.fast can deliver cellular data over copper lines at speeds of 150 – 200Mbps, potentially laying the groundwork for 4G and 5G network deployments based on G.fast.