Vodafone Idea tests revolutionary approach to network deployment

Vodafone Idea, the country’s third largest mobile operator, has started deploying a revolutionary new approach to creating its cellular network, but has also encountered several challenges in doing so.

The company is experimenting with the so-called ‘open radio access network’ or Open RAN technology, which is a break from the traditional one-box technology promoted by the likes of Nokia, Ericsson, Huawei and others.

An open radio network is created using off-the-shelf components, such as intel or AMD-based motherboards, on which the company installs an open operating system and software sourced from vendors to add different functionalities.

Under the traditional approach, a cell-site was constructed around a ‘box’ provided by Ericsson, Nokia, Huawei and so on. The operator’s job was simply to connect the box to a power supply and and provide it back-end connectivity using a fiber network or microwave.

The box vendor — Ericsson, Nokia and so on — would then control and manage the box deployed at the cell site remotely using the network.

However, a clutch of smaller companies have sprung up — such as ArrayComm, Altran, Wind and CIG — which offer software to convert an off-the-shelf motherboard — running a specially designed Linux operating system — into a feature-rich base station.

Most of them design their software for deployment on Open Radio Access Network or O-RAN — an operating system for base stations curated by the Linux Foundation.

The computing board, on which the O-RAN software is deployed, can be built around x86 chips (such as those from Intel or AMD) or more commonly, around an ARM chip, such as those from Qualcomm, Mediatek or Broadcom.

This is similar to replacing a Netgear or TP-Link home router with a Raspberry Pi board with OpenWrt operating system running on it.

While such a move may not make much sense for someone who has just one router at home, it makes a lot of sense for an operator like Airtel or Vi, which has lakhs of routers (base stations) spread across the country.

Not only does such a move reduce costs by increasing the choice of hardware and their re-usability, it also helps these companies do things to their base stations that could not even be conceived of under the earlier ‘closed box’ model.

Since the base station is open now, telecom operators can also install many other software on these, including those that help with intelligent and automated traffic control, as well as the delivery of value-added services to customers.

In other words, the opening up of the base station can make the ‘network edge’ as smart and intelligent as the network core, instead of being just a traffic router.

Rajesh Singh, executive vice president at Vodafone Idea and in charge of the company’s radio planning and network-related R&D efforts, spoke at length about his company’s experience in trying to move from a closed-box model to an open-board model as far as Vodafone Idea’s network was concerned.

At this year’s India Mobile Congress, Singh highlighted how the ‘edge’ of the network has not benefited from technologies such as cloud computing and analytics, while both the core network and the IT functions such as billing and provisioning benefited tremendously.

The following are the excepts of his talk on the subject:

“O-RAN is one of the most awaited things in radio technology. We have talked about cloudifications in other domains like core, ITES.. Somehow, the RAN could not leverage cloudification as it has not moved to a different architecture. It remains in the one-box, proprietary kind of [model] with hardware and software being delivered by the OEM [equipment vendor].

“Somehow, the OEMs — the Tier 1 supplier — never has encouraged [others], I say, to move out of this whole proprietary solution, though they’re very good at it, and they could very well control it, being delivered in one box.

“Somehow, [for the] operator also, time-to-market was a critical thing and they would just take this one box, put it on the network and offer the services and go on. The support is also provided by the same OEMs. 

“Earlier also, there have been discussions around CRAN. It has not taken off. I don’t see much deployment of CRAN.

“Then came virtual RAN. But virtual RAN was also supplied by the same OEMs, so they failed to leverage cloud infrastructure, because you couldn’t go to the edge level, and it remained near the core domain itself.

“So we could not leverage whatever transformation was happening in COTS [commercial off-the-shelf] hardware, moving from proprietary to off-the-shelf standardized hardware of GPP [general purpose processor] or x86 based. That leveraging was missing.

“ORAN can now accelerate this path as splitting and disaggregation has started. Hardware, software is now from two different vendors. You can choose the best of the breed, and you can deploy this in the network, wherever scaling is required.  

“At the same time, you get flexibility. You can choose your hardware from the best of the vendors, and you can have software that is best its class. And maybe, you can customize as well, because when the proprietary solution was coming, it was not possible by the operator to really do any customization to their network, because it was totally locked by the suppliers. All operators were having the same boxes, same services.. so no differentiation was possible…

“So [in the new setup], you can differentiate your services, you can operate with APIs and third party partners, you can have lots of innovations where you can go to edge services, and close to the customer. A lot of enterprise and SME services can be delivered from the edge, rather than delivering from the DRAN architecture. 

“Another thing is that as the data demand is going through the roof and the growth of data consumption is not stopping at all.. and per head consumption is also one of the highest in the market. So, how [do we] bring down the cost of the operation? 

“We see Open RAN as a driver to drop (sic) the cost. As we open the ecosystem [more], more suppliers will join in this ecosystem, competition will go up and it will commoditize the radios, and your capacity, which was locked simply to one box, it was nailed to one site — this will go towards the pooling efficiency you can achieve. You don’t need to deploy to all the sites, which currently is currently the practice, but you can — as and when demand rises — give the service and capacity to that particular site, instead of giving nailed capacity to every site in a uniform way. That flexbility will come.

“Another important aspect will be the system integrator. Because when you choose your software from x vendor and hardware from y vendor, you have to see that they get integrated properly and can deliver the service. The proprietary boxes [on the other hand] are very mature, and very very robust in their nature and that’s how reliability and resilience is ensured in the network…

“The other part is how to optimize opex [operating expenses]. We also see OPEX advantages. Currently, we are using very complex technologies, whether it is 4G, Massive MIMO, multi-beam forming — all these are very complex technologies.

“The optimization challenges are getting very very critical. So [the question is] how to optimize better and achieve automation. Once you have software driven radio, with open source software and a multi-supplier market where suppliers are competing with each other, we see AI and ML can be embedded in this one, and your optimization challenges can be smoothed out and you will not need manual interventions, and it can be totally automated. You can really enrich your software in the open RAN architecture…

“Even the hardware, you are splitting [into] CU [centralized unit], DU [distributed unit] and RRHs [remote radio heads], your space and power requirements will reduce at the tower. So, your energy costs and loading costs get reduced. So you get further opex benefits when you adopt the ORAN architecture.. 

“At the same time, there are other, practical challenges…In current 4G network, we have done some experiments in the FDD layer in the B3 band. We have tied up with a few vendors in the market, and we have deployed [their products] on the network, and we have seen the real KPIs compared with existing vendor products. 

“What we see is that current vendor products are very mature, and we have struggled to match those KPIs for quite a long time, as the traffic demand and the capacity demand is very high from the first day itself, since it is not a green-field network. The demand on KPIs is very high on the first day. You can’t wait for that to get settled.

“But here [in ORAN], the system integrator is there, the hardware is there, the software is there; so, we have seen that integrating all these together and matching with vendor supplied, proprietary, single-box product — robust, hardened — there were some challenges in matching the KPIs. Either it is [lower] throughput or [higher] drop rate. To add to the complexity, we have VoLTE on top of it.

“So, with all this, we’ve seen it is a challenge [to match the KPIs], but we have worked closely with the new vendor, and we could manage. But it took almost 4-5 months to come up to that level of maturity that was offered by the existing proprietary tier-1 vendors’ products.

“We also see some challenges when we talk about brown-field [existing] networks…What we seen in ORAN is that generally, they are offering [only] the 4G layer. It is very difficult to have backward compatibility with 2G…

“2G demand is still there in the market. If you are splitting 2G and 4G [base stations], then you lose the SRAN [single RAN base station] advantage, which the current boxes are providing to us.

“The current tier-1 vendor can manage 2G-4G very well in the same box with higher capacity. But we see that the Open RAN supplier has that challenge — that they cannot manage 2G on top of their software stack.

“We are working with them so that we can [develop] at least some backward compatibility, so that we can plug it in, in the Indian market, and we can serve the brownfield market, and not only the green-field network…

“A little maturity is coming in the B3 (1800 MHz band) product, but as soon as you move towards the TDD layer and Massive MIMO, again, O-RAN vendors are finding some difficulty in offering those products. We have to see that all bands can be covered. We have to operate all bands.. products have to come in all the bands, rather than sticking to the B3 band, which is the really mature kind of product.”