22 Mar 2021

A brief history of Open RAN

As lots of people point out, Open RAN is an idea or a movement. The fact is that RANs (Radio Access Networks) have consistently been closed, in that a mobile network operator (MNO) had to buy every part of their RAN from the same vendor for it to function. There have been several attempts to change this so-called vendor 'lock-in'. If we look back to when the WCDMA 3G standard was being defined in the 1990s, WCDMA split the RAN into two major components; the (interestingly named) "Node B", the physical layer of the base station, and the Radio Network Controller (RNC). One RNC controlled a group of Node Bs, and the interface between them, known as the IuB interface, was supposed to be standardised so that MNOs could mix and match RNCs and Node Bs. Naturally, it wasn't in the vendor's interests to do this, so there was no shift or momentum to implement this standard.  

The drive to a more open market

Several years later, a Home Node B concept evolved, which became known as a femtocell (now known as a small cell). A group of six (largely UK-based) companies formed an organisation known as the Femto Forum, which later changed its name to the Small Cell Forum (SCF). Famously, at one Femto Forum meeting, Vodafone representatives were alleged to have refused to let the vendors leave the room until they had agreed on a standard interface between the Home Node B and the security gateway, leading to the IuH interface being born. In this case, the MNOs had the upper hand because startup companies were primarily developing femtocells. It was in these startup companies' mutual interest to create an open market to compete with the large incumbents.  

LTE (4G) removed the need for this interface as the RAN functions were all aggregated into one unit, the eNodeB (for evolved Node B). As the 4G era progressed, challenger small cell vendors emerged. These smaller vendors found that the interface between the eNodeB and the core, the S1 interface, was pretty standard, and there was good interoperability. The X2 interface, which allows eNodeBs to talk to each other (for example, during handover between cells), was not standard. Worse still, there were problems with the management interface.  

The management interface of any telecommunications equipment handles monitoring, maintenance and software function upgrades. MNOs look for efficiencies and need a unified management view of their network, not requiring staff to look at separate and disparate screens for different equipment pieces from varying vendors. Without a unified management interface, it made it difficult to mix and match equipment from alternative vendors. Most MNOs managed to get around this and have more than one RAN vendor by dividing their territories into separate parts (for example, one for the north and one for the south in the UK) and use different vendors for each of these regions.  

During this time, competitive pressures led to a considerable reduction in the number of RAN vendors across the globe (that's another story). This choice-of-supply cull led to the MNOs feeling that they had little choice but to accept whatever products the few remaining vendors offered at a price they wanted to charge. Although this is somewhat of a distortion of reality, it led to the formation of two operator-lead organisations: the C-RAN Alliance, whose leading light was China Mobile, and the XRAN Forum, based around AT&T. After a while, it became apparent that the two organisations had a lot in common, so they agreed to merge in 2018 to form the O-RAN Alliance.  

The need for standard interfaces

Although 3GPP defined a hardware specification for a gNodeB (a next Generation Node B) in 5G, a major tenet of the Open RAN movement is that much of the gNodeB can be realised in software rather than special-purpose hardware. Advances in sheer processing power have enabled this shift. Advocates of Open RAN argued that if they split the RAN into component parts, roughly corresponding to Layers 3, 2 and 1 of the protocol stack, NMOs could buy standard computing hardware and purchase software components from various vendors. They would include both traditional RAN vendors and new entrants, thus opening up the RAN to competition. 

However, standard interfaces needed to be defined between the different parts and vendors. Whether providing software or hardware, products would need to conform to these specifications, and the organisations within the Open RAN movement have begun defining these interfaces. 3GPP initially defined the first of these, known as the F1 interface between layers 2 and 3 of the protocol stack. It includes the layer three functions of many gNodeBs concentrated into a Central Unit. The O-RAN Alliance and, specifically for small cells, the Small Cell Forum have defined additional interface specifications. These include management interfaces, Open Fronthaul by the O-RAN Alliance with the intelligent parts of the gNodeB concentrated into the O-RAN Alliance-defined RAN Intelligent Controller (RIC), and the nFAPI specification from SCF for the layer 2-layer 1 interface. 

At the start of 2021, these interfaces have reached a maturity level where equipment from different vendors could conceivably interoperate. However, the interoperability demonstrations to date have been somewhat proprietary. Though we are indeed at the cusp of change with momentum ever building, will the Open RAN movement fully materialise? The question remains whether NMOs will opt to use Open RAN equipment in sufficient volumes to build a viable industry or whether they will be lured back into the comfortable world of vendor lock-in? 

Peter Claydon is President of Picocom. Peter Claydon is president of Picocom. He has a career spanning 35 years in the fields of semiconductors and wireless, ranging from avionic systems through video encoding and digital TV through to 3G, 4G and 5G cellular systems. He was a founder of UK, Bath-based company Picochip, where he oversaw the creation of the first chips used in small cell cellular base stations that are in-use in their millions throughout the world. Peter also worked on the development of 4G base stations at Airspan Networks and as an applications consultant at UltraSoC. Immediately prior to joining Picocom, he was the project director of the AutoAir 5G testbed.You can follow Picocom on Twitter and LinkedIn

To read more from #DiversifyingTelecoms Campaign Week check out our landing page here.