Intel positions itself for 5G tipping point
DAN RODRIGUEZ
Corporate VP and GM of the Network Platforms Group Intel
As a key technology leader in the industry, Intel explains how it contributes to network transformation efforts that are underpinning successful 5G deployments, and explores what lies ahead for communications service providers (CoSPs).
The coronavirus pandemic has placed mobile and fixed telecommunications networks under huge strain during 2020, with video streaming, gaming, video calling, as well as education and collaboration tools, all competing for capacity at the same time.
It’s fair to say that the networks have performed better than perhaps anyone would have expected. As explained by Dan Rodriguez, corporate VP and GM of the Network Platforms Group at Intel, networks have proved to be “incredibly resilient” in the way they are designed.
“We have seen that most networks have been able to adapt. And I believe a large part of that is owing to the network transformation that we drove as an industry,” Rodriguez said. “Our customers and partners say that the world is really viewing the network as an essential service.”
When you consider that this has been possible in the 4G era, the opportunities that 5G technology will open up are even greater. Rodriguez said consumers will benefit from a better quality of experience because of the lower latencies and higher bandwidth of new 5G networks. As for enterprises, “service providers are really expecting to deliver better outcomes for businesses,” he noted.
“It’s a time for all of us to continue to move forward with plans to really deliver a world-class network on 5G,” Rodriguez added.
Becoming more agile
5G in still in the early stages of deployment, especially standalone 5G core networks, but rollouts are gathering pace. Spectrum auctions are also now taking place following delays caused by the pandemic.
According to the GSMA, 107 operators in 47 markets had launched commercial 5G networks by the end of September 2020. Around 121 operators in 35 markets had acquired 5G-enabling spectrum in 35 markets, and trials are underway by 217 operators in 100 markets. By 2025, 5G connections are expected to reach around 1.8 billion, with 410 operators offering 5G services in 123 markets.
The real success of 5G will depend on the move towards more software-defined, agile and scalable networks, “utilising some of the tools and technologies that have been honed in the cloud and the data centre,” Rodriguez said.
In one example, Intel partnered with Korean operator SK Telecom to “maximise the 5G core network performance, utilising our Xeon scalable processor as well as our Ethernet adapters. In this case, we were able to tune up that core network, resulting in a 78%-88% reduction in latency and jitter”, noted Rodriguez.
He added: “Think of consumer services that are enabled by 5G. Think about cloud gaming. Think about video streaming: with lower latency and jitter, you’re going to improve those overall experiences. And then for enterprises, think about immersive, visual services, like augmented and virtual reality: you can see that a reduction in latency and jitter is going to have a meaningful impact in terms of the effectiveness of new immersive use cases.”
Looking further ahead, Intel believes that the next wave of transformation will benefit from a “perfect storm” of 5G, edge buildout, and the pervasiveness of artificial intelligence (AI).
“This is really driven by the need to deliver network services as well as new AI-based edge services across multiple network locations, across multiple vertical markets. These vertical markets include everything from industrial, retail, education, healthcare, smart cities, smart venues and many more,” Rodriguez said.
The core of the matter
Intel plays a critical role in ensuring the success of future 5G networks, both because of the work it has engaged in over the past ten years and its current technological advances.
“Our role is to partner with the overall ecosystem and bring commercial solutions to the market that really support the aims of 5G in the spirit of network transformation,” said Rodriguez.
In the network core, Intel started on the network transformation journey about a decade ago, working in partnership with the entire industry ecosystem. In 2013 Intel kicked off a network functions virtualisation (NFV) project with European standards body ETSI, in collaboration with CoSPs, “and that led to all sorts of proofs of concepts and trials to deliver cloud technologies such as virtualisation in a standard way to the network,” Rodriguez reflected.
In turn, that fostered a huge amount of experimentation in the industry, “to really move the market towards software-defined, agile and scalable networks,” the Intel executive commented.
“And the great thing here is that the wireless core network has really started to adopt this significantly over the last few years and it’s on track for more steady gains,” he added. For example, this year alone analysts forecast 50% of core network deployments to be transformed into virtualised network servers. Rodriguez also pointed to analyst projections that more than 80% of all core network deployments by 2024 will be virtualised.
“We are definitely seeing CoSPs really increase the pace as they move to a 5G standalone core from a non-standalone core, and we expect that to continue,” he said.
Going to the edge
In the radio access network (RAN) area, Rodriguez said Intel expects to see a mix of traditional and virtualised RAN solutions for some time. CoSPs typically operate across very diverse environments and geographies and will require different deployment options for some time to come.
“We are starting to see a lot of traction in the marketplace that really shows CoSPs and the ecosystem are driving the network all the way to the RAN to become much more virtualised, to use more cloud technologies to be able to have that agility at the far edge of the network,” he said.
Rodriguez pointed to examples of operators that are already deploying virtual RANs, such as Rakuten Mobile and Verizon.
Intel is also partnering with satellite TV operator DISH on the deployment of virtual RAN and open RAN. Intel’s 5G infrastructure technology will create the foundation for a greenfield 5G network at DISH, including the Intel Xeon scalable processor, the Intel Ethernet 800 Series network adapter, the Intel vRAN Dedicated Accelerator ACC100 and Intel FlexRAN software reference architecture.
Furthermore, China Unicom has announced the launch of large-scale edge cloud field trials in collaboration with Intel, as well as Tencent. The China-based operator ultimately plans to provide differentiated network capabilities and services in vertical industries using its EdgePOD platform. The platform is based on servers using the Intel Xeon scalable processor family, along with OpenNESS, DPDK, SR-IOV and other optimisation technologies.
“That’s a good example of a customer saying I’m going to put compute and a multi-cloud environment right at the edge of my network that can easily tap into that 5G network to deliver some amazing services across a wide variety of verticals,” Rodriguez said. “We are definitely expecting to see many more of these examples coming into play over the next few years.”
Reaching a tipping point
Intel is also working hard on delivering new technologies, including software tools, to support this new virtualised, agile 5G era.
For example, its OpenNESS edge computing software toolkit and OpenVINO developer toolkit focus on fostering open collaboration and innovation. OpenNESS enables highly optimised and performant edge platforms to be able to on-board and manage applications and network functions with cloud-like agility across any type of network. Rodriguez pointed out that OpenNESS is at the heart of the cloud environment that powers Rakuten’s 5G network. OpenVINO offers a development environment for inference, computing vision applications and hardware acceleration, which is driving innovations in edge AI acceleration.
“We obviously need to bring a full suite of feature-rich silicon,” Rodriguez said. “We’re investing in everything from CPUs, to FPGAs and ASICs, and a host of other platform ingredients such as ethernet, to really bring a wide variety of solutions together for the various CoSP networks to be able to hit their performance and power as well as their cost goals. Additionally, the general-purpose nature of our silicon also supports that movement towards network transformation, being more software-defined, being more agile, and more scalable.”
Rodriguez also pointed out that Intel is spending “a ton of effort” on really creating a rich set of software and tools, “allowing our customers to get the most out of our silicon and also doing it in a way that allows them to create a really open and standards-based network where they have plenty of agility and scale.”
At the same time, of course, Intel is participating in standards bodies from ETSI through to the Linux Foundation and the Telecom Infra Project (TIP) to help drive virtualisation and open network agendas.
It’s now up to the CoSPs to ensure they will build 5G networks that support today’s needs and the future demands of both enterprises and consumers.
“We are really at a pivotal moment,” said Rodriguez. “We are at a tipping point in the industry. We are seeing the new wave of network transformation driving a rapid transition to virtualisation and cloud-native technologies as well as edge computing.”
In his view, it is essential that CoSPs “deliver a diverse set of use cases for consumers but also for a wide variety of enterprises across multiple vertical markets”.
“It is critical that CoSPs build a network from the ground up with the right flexibility, scale, and agility in mind, and this will help them not only better manage their total cost of ownership in future but also deliver all sorts of amazing experiences to consumers and businesses and set them up for long-term success,” Rodriguez concluded.