In late February, during US president Donald Trump’s trip to India, he reportedly asked Reliance Industries chairman Mukesh Ambani if his mobile operator Reliance Jio had plans for 5G. Ambani told him that not only is Jio planning to launch 5G, it will be the only 5G network on Earth “that doesn’t have a single Chinese component.”
When Ambani said that, the assumption was that he meant Jio would not buy 5G gear from Huawei or ZTE, but would instead stick with Samsung, which also makes 5G equipment and is Jio’s current 4G supplier. Jio listed Samsung as a 5G partner when it submitted its 5G proposal to the India’s Department of Telecommunications (DoT).
So imagine everyone’s surprise when, earlier this month, the Economic Times reported that Jio would forego vendors altogether and use its own in-house 5G technology instead.
To which analysts and telecoms industry observers around the world said: What?
From the ET report:
“We have now developed everything end-to-end around 5G technology. We are more scalable than these vendors and are fully automated since we have our own cloud-native platform. In 5G, we will totally be self-sufficient,” the person told ET.
It should be stated at this point that ET got this from an anonymous source, and that Jio hasn’t confirmed anything. But the implication is that Jio has literally developed every component it needs to build and launch a complete 5G network, from the base stations and antennas to the core network.
Which would be something.
Is it true? No one knows except Jio. Is it plausible? More than you’d think.
First of all, Jio began life as an all-IP 4G network, which means it has no legacy circuit-switched 2G/3G gear to deal with or maintain. Because it’s designed for IP (or “cloud-native”, as ET’s source puts it), most everything is done in software. And, as it happens, Jio has a lot of mobile software expertise at its disposal. One of Reliance’s subsidiaries is 4G software firm Rancore Technologies, which bought acquired US-based software company Radisys in 2018.
This is perhaps the key to the puzzle, because the telecoms sector in general is moving towards an era where telecoms operations are done in software that’s decoupled from the hardware. This is after all the whole point of SDN and NFV. Moreover, as operators undergo digital transformation, more and more of them are moving some or all of their B/OSS systems to public-cloud providers like AWS, Azure and Google Cloud.
As for 5G, according to a new paper from Rethink Research [PDF], the evolution roadmap for 5G eventually culminates in a cloud-native 5G core network whose functions are disaggregated and deployed in containers as microservices, and then a cloud-native radio access network (RAN), at which point operators will ditch pricey proprietary vendor hardware in favor of specialized open-source 5G software running on commodity hardware.
There are several industry groups working on this ‘Open RAN’ concept, including Facebook’s Telecom Infra Project and the operator-led O-Ran Alliance (both of whom just signed a deal last month to coordinate their activities). The basic idea is to turn base stations into white boxes running custom software. This wouldn’t cut existing 5G vendors out of the picture – it would simply force them to shift their business model from hardware to software, thus enabling open multi-vendor RAN environments.
While Open RAN is believed to still be at least a year or two away from commercial reality, Japan’s Rakuten Mobile – which, like Jio, is a greenfield cloud-native mobile network –claims to have already done it. Earlier this month, Rakuten Mobile announced it had successfully deployed a fully virtualized, cloud native, open RAN (albeit six months behind schedule), and will use it to launch 4G services in April, and will upgrade it to 5G in June.
The patent problem
So it’s possible this is what Jio more or less has in mind – in which case it arguably has the software skills to skip the 5G vendors and take the DIY route. (As for the hardware, the ET report claims Jio has designed the necessary hardware that can be manufactured by third parties.)
On the other hand, Rakuten Mobile didn’t skip vendors in favour of in-house tech – it contracted a range of vendors for equipment, software, and services. For example, Altiostar (which Rakuten owns) and Nokia are reportedly supplying the 4G antennas, while NEC will build the massive multiple-input multiple-output (MIMO) antenna components needed for 5G. Airspan, Cisco, Intel, Mavenir and Qualcomm are also listed as vendors.
The other thing about 5G equipment – especially when it comes to antennas and radios – is that it’s highly specialized and sophisticated technology. More to the point, it’s highly patented.
When Vietnamese operator Viettel announced last year – and again this past January – that it was going to not only deploy its own 5G hardware and software but also sell it to other operators, a number of analysts expressed doubt. Their main objection, according to the BBC, was that 5G isn’t the sort of technology you can slap together in your basement in a few months, and even if it was, you’d have to license the necessary technology from the very vendors you don’t want to buy equipment from, which would not be cheap.
In any case, we’ll find out sooner or later once Jio launches its 5G network. And it will probably be later. The DoT had hoped to auction off 5G spectrum next month, but last week it decided to delay the auction to August for an all-too-familiar reason: its spectrum prices are so high that India’s operators – who are all running on razor-thin margins in a hypercompetitive market – can’t afford it. The general expectation is that India may not see 5G services until at least the middle of next year, or even until 2022.
Plenty of time for Jio to gets its gear ready, perhaps.