Based in Ireland, Brendan Farley is managing director of Xilinx EMEA and vice-president for wireless engineering with responsibility for the development and roll-out of technologies that will support the deployment of 5G infrastructure.
Appointed to the position last year he has to manage teams and a complex R&D process that spans Ireland, the UK, Germany, US and India while ensuring that all the company’s key functions, whether that’s R&D, sales, marketing, HR, IT, finance and legal are working effectively together.
“Our work is focused on developing the infrastructure that’s needed to deliver 5G and that means working very closely with our customers and partners, as they look to deliver new applications and services,” Farley explains.
“Xilinx’s products are used in cellular base stations, and the roll-out of 5G has proved a significant opportunity for the company and much of the R&D activity takes place here in Ireland.”
Prior to his appointment Farley held a number of senior engineering management roles within Xilinx, and before joining the company had worked for many years in the Irish technology sector.
“It’s been an interesting few months and my appointment, which coincided with changed working practices as a result of the pandemic, has meant that it’s been even more important to stay in touch with the team and keep people motivated and informed.
“Likewise with our customer engagement we’ve probably been far more active using tools like Zoom, but I think there’s now a real desire to get back to normal and re-establish those personal relationships that are so important in this business. However, we don’t want to lose the benefits working online has brought, whether that’s breaking down old formalities of lots of people having to turn up on site to make decisions or simply enabling decision-making to be faster.”
Turning to 5G and the roll-out of the infrastructure necessary to support it, Farley says that the acceleration in the commercial deployment of 5G is driving the market forward.
“The commercial roll-out of 5G has certainly picked up and as a consequence we’re seeing increased investment in network infrastructure in order to meet the needs of 5G users. Xilinx has been front and centre here, helping to upgrade networks across China, South Korea and in North America.”
The strength of Xilinx’s portfolio lies in the programmability and adaptability of its FPGAs and SoCs.
“Specifications are constantly changing,” Farley points out, and “along with the features that need to be supported. It’s difficult to keep up with these changes if you’re designing with ASICs, so the flexibility that our adaptable SoCs and FPGAs provide has made them far more suitable.”
A phased roll-out
According to Farley the roll out of 5G will need to be phased.
“The first phase is about upgrading the infrastructure from 4G to 5G in order to support traditional operator business models. Our technology is being used to increase capacity and reduce latency and that’s happening now – and we’re playing very strongly in that.
“For operators though that’s not enough. They will not consider 5G a success until they are able to develop new revenue streams from new verticals – and that will require heavy investment. But that’s essential and that’s what we’ll see in Phase 2.”
Farley suggests that Phase 2 will see new revenues from new applications which will be driven by the operability and integration of various OpenRAN solutions.
“This whole process is closely tied in with OpenRAN, one of the cornerstones of which is the separation of hardware and software, particularly at the edge – in other words the base band next to the tower. At the moment the use of ASICs is widespread and there’s a focus on proprietary hardware. We are now seeing a move to the virtualisation of the base band with software doing most of the signal processing that was traditionally done by an ASIC.”
The advantage of OpenRAN and the use of software is that it is now possible to push third party applications on to servers which will be able to process data in real-time.
“That will open up a host of opportunities for new applications,” Farley suggests. “Just look at the automotive space and the concept of cellular vehicle to everything – it you have low latency all the traffic will be passing through a base station tower to another end user. That will make it possible to implement augmented reality, for example, in the form of holographic navigation systems. And, if you can connect cars to a 5G tower, you can connect traffic lights and pedestrians and if you are able to deliver latencies of under 50ms, it’ll be possible to significantly improve road safety.”
There is, however, still some way to go before we are able to deliver these new services. Many would argue that we need to be careful about not over-promising what is deliverable and being realistic about possible timelines, especially when it comes to 5G standalone services.
“For sure, I would agree with that,” says Farley. “In terms of a basic route map we need to upgrade the infrastructure first and then virtualise the base band and that has started to happen. Samsung has already demonstrated that it’s possible to get the same level of performance from a virtualised system, but not everything can be run in software. Some functionality is only possible through hardware.
“This is a vision that will be captured in new 3GPP standards as they emerge but we do need to be able to demonstrate increased 3X capacity with the existing business models; then we need to virtualise and disaggregate the network, especially the base band with some acceleration from FPGAs, and then we need to engage with the verticals, whether that’s factory automation or autonomous vehicles.
“We need to get buy-in from those new verticals and moving to ‘server based architectures’ will be critical. For factories that could result in a move to a wire-less factory. That will require acceleration at the edge, but once the software platform is in place people will start to see the opportunities.”
A successful deployment, however, will depend on an ability to provide low latency especially, for example, in a factory environment that’s looking to operate automated processes, robots and machine vision applications.
“All of this requires the 3GPP to reach out to these verticals, to see where new value can be created and what the new opportunities are for operators,” says Farley. He continues, “The key thing is that all of this remains a moving target.
“Whether we are talking about standards or features, something is being added every year and that’s why the programmability and adaptability of the devices that Xilinx provides are critical when it comes to 5G.”
