Open RAN and “White Box” Radio

Certainly very few people with an interest in mobile networks are unaware of the advent of Open RAN. Sometimes, though, the concept gets a little murky with all the discussion of “disaggregated software” running on “commercial off-the-shelf hardware” and using open APIs and even “white box radios”.

The idea of “white box” recalls ATT’s Domain 2.0 program that heralded virtualization in telecommunications networks. (For example, see the ECOMP paper from 2016.) White box switches and routers are examples of “hardware/software disaggregation”, an inelegant phrase that refers to the now accepted practice of generic hardware that will run forwarding software from any of a number of vendors. Generic hardware by definition can be built by anyone, and with the right drivers supports anyone’s software, leading to competition among hardware and software vendors for space in the enterprise rack.

Sounds like a no-brainer, especially since the concept has proven itself in practice. The ORAN Alliance even has a workgroup dedicated to “white box radio”. (See for example WG7 on this page.) Workgroup 7 publishes reference designs and architecture specifications for base station radios, as well as for servers for RAN applications.

But this is not what it seems.

First of all, WG7’s reference designs lack a hardware abstraction layer. This is not by accident. There is simply no intent to define an internal API for radio hardware, à la a PC BIOS or Open Compute’s Switch Abstraction Interface. ORAN defines an information model that applies to the radio, but this describes information elements carried over fronthaul between the radio and its controlling server, not how the radio hardware is represented internally to the radio software.

In other words, even with WG7’s reference designs, the radio software is proprietary to the radio. The radio product is in fact a black box. It has an open API to the rest of the network, but the network operator does not — can not — control the software.

But what if hardware vendors follow the ORAN reference designs? If all hardware designs for a given application are the same and openly available, wouldn’t that make a hardware abstraction layer unnecessary?

Perhaps. But then the only way for an OEM to differentiate competitively would be on price, which is not an attractive proposition — especially for a radio-savvy OEM who has the ability to optimize for particular needs.

The same observation applies to Facebook’s Evenstar program. And in fact, there is only one hardware vendor bringing a single 4T4R radio on band 3 to market under this program, MTI.

Now, MTI is a capable OEM. But singly they hardly qualify as a radio “ecosystem”. And the ecosystem is what open RAN is all about.

Mavenir for one certainly recognizes this. It recently announced the creation of a business unit devoted to radio hardware. In an interview with Fierce Wireless, Mavenir’s CEO stated, “open RAN won’t be successful unless there are radios that suit the needs of many operators who all have different spectrum and different requirements.” (Article here.)

For us to address a global market, we would need over 100 radios. We cannot do it ourselves. Our goal is to give reference designs to people and other companies so they can build radios that will create an ecosystem.

A worthy aim.


Mavenir will open source the reference design for its Massive MIMO radio after it’s completed. ‘This is our own internal IP,’ said Kohli. ‘To make something open source, you have to own it. We are first building it ourselves and then will make it available. We’ll also build the radio software and make that open source as well.’

This is quite ambitious, although quite vague as well. It leaves one of the biggest questions unanswered: intellectual property rights.

Specifically, patents. Although so-called “essential patents” (essential to implement an industry standard) are generally associated with software, some of that software resides in the radio, particularly massive MIMO radios. Even if it’s open source, a patent is a patent, and the patent holder gets paid.

But the hard part with radio patents is their sheer number and diversity. So the question of IPR becomes who tracks down all the relevant patent owners and negotiates licenses. If Mavenir plans to implement their own patent pool with pre-negotiated licenses, I don’t know why they don’t shout it from the rooftops. It would be a game-changer.

Because this is how the game is played with handsets, modems, C-V2X, etc. Many patents are wrapped up in silicon, and the OEM pays the price in a simple transaction with the silicon vendor. Other patents are available through patent pools like Avanci’s. But these are for user devices. On the network side, i.e. the RAN, there is little to no relevant IPR wrapped up in silicon (apart from small cells), and no pools that I am aware of.

Radio for mobile networks is hard enough. Many bands, FDD as well as TDD, many operators with many disparate needs, multiplied over many MIMO levels. And rumor has it that MTI really struggled with their single-band 4T4R for a time, which is a very basic radio. Without a turn-key solution to patents, it’s hard to envision an ecosystem of small, scrappy radio players that open RAN evangelists seem to expect. Rather, at the moment it seems much more plausible that Fujitsu, NEC, Nokia, Samsung, and eventually Ericsson will become the main players.

Unless Mavenir has a pool up its sleeve.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: