All players in the business of wireless networks — network operators, and their hardware and software suppliers — have to have an opinion on Open Radio Access Networks, i.e. networks defined by the ORAN Alliance.
As 4G is upgraded, as 5G is deployed, is ORAN the way to go? Or not?
That kind of opinion.
The airy promise is that networks based on ORAN will be just as good as proprietary networks engineered by the likes of Ericsson, Nokia, and Huawei, but at a substantially lower cost — low enough to justify what might be a risky investment. But you have to wonder how this can be possible, when Ericsson and Nokia each manage to eke out an operating margin of just 4% or so, or when Huawei is reported to spend more on R&D annually than all their competitors combined.
The “savvier” view is that a greenfield ORAN network doesn’t have to start out as good as the best that Ericsson can deliver. It only has to be ‘good enough’, so long as it’s cheaper. Over time, the ORAN ecosystem of hardware and software vendors, being more competitive (because, well, “open”), will get better and better. The old-line OEMs will either join the movement or wind up on the scrap heap.
More than a few people even think that ORAN software will all become open-source, all hardware will become commoditized white-box radios and servers, and building a radio access network will eventually become no more exciting than building another data center.
Radio access networks have two problems to solve: Coverage, and Capacity.
Coverage is the comparatively easy part. When you deploy enough radios, everyone is covered. It’s a straightforward matter of spending money and performing well-understood site engineering. Not always easy, mind you. But well understood.
Capacity is different. Of course you have to keep spending money. But even with that, capacity comes down to managing spectrum and working with interference.
There’s a reason why we all carry “cell” phones and not cordless phones. Our wireless experience is provided by a network of cells, each cell served by a base station. Sound elementary? But cells have edges, and the edge is not just the perimeter of the cell, it’s where cells overlap. And this is the zone where interference sets the rules of the game, where phones are within range of two or more base stations operating on the same licensed frequencies, where a base station “hears” the transmissions of phones that are really being served by other base stations on top of the phones served locally. And there are probably about as many wireless users in these zones at any one time as not.
Once all of the radio hardware magic is in place, then managing the carriers and frequencies and timeslots and MIMO precodings, and all of that, is a software problem. And there is a key piece of this software called “the scheduler”.
Now, the ORAN Alliance can publish APIs for interfacing to a scheduler, but the scheduler itself is a proprietary engine. There is no standard scheduler. It’s algorithms are a key ‘secret sauce’, and old-line OEMs like Ericsson invest many R&D dollars evolving their scheduling algorithms along with evolving radio technologies, maybe adding a little machine learning along the way. The result is a piece of intellectual property they guard very, very jealously.
There is of course much more to the radio access network, probably many millions of lines of code performing many essential functions, and companies like Mavenir and Altiostar and Radisys undoubtedly have capable product portfolios. But compare it to TikTok. Here, too, are millions of necessary lines of code performing necessary functions like database management and a user interface, most of which could be replicated by any competent software shop. But the design of the recommendation engine is TikTok’s secret sauce. It’s why TikTok is a household name.
The scheduler is an example, maybe the best example, of why commodified hardware and open-source software may go only so far in 4G and 5G networks.
I have no doubt that an ecosystem of ORAN vendors will eventually grow up with products that can be assembled to deliver wireless coverage. The question is, whose products will deliver the capacity needed for congested urban networks — because urban networks are where the most users are found and the most data demanded.
In other words, where the money is. If you’re a network operator.
Still, open APIs are a compelling force. Even Nokia is coming around to embrace them, and for operators this means options (and competition) in areas of their networks where cost is paramount. Maybe rural areas. Maybe in building out the 5G coverage layer.
But where the driving need is capacity, i.e. the number of users who can be served in a given area with an acceptable data rate, over a fixed amount of expensive spectrum — here an operator can’t pick and choose on cost alone, and the actual viable software and hardware vendors may look a lot less like an ‘ecosystem’ and much more like an exclusive club.
To be continued!