With one notable exception, commercial 5G networks use a non-standalone architecture wherein new 5G NR radio infrastructure is connected to an Evolved Packet Core that also supports LTE data transmissions. Virtualization of EPC functionality is common but stops short of the cloud-native core network needed for a standalone 5G implementation. With non-standalone 5G, the emphasis is on delivering enhanced mobile broadband to consumers. To enable the full feature set of 5G, including reduced latency, high reliability and support for massive numbers of connected devices, operators need to adopt a standalone 5G architecture.
Table of contents
Operator focus: T-Mobile
Operator focus: Rakuten Mobile
Operator focus: Vodafone UK
Leveraging latency requires decentralized compute
New slices, new services
From standardization to commercialization, 5G has largely been ahead of schedule. The transition to standalone is ramping up too; in the U.S. T-Mobile has activated standalone for its low-band 5G network and Verizon and AT&T plan to make the jump in the coming months. While standalone brings numerous benefits, perhaps chief among them a clearer path to enterprise service revenues, the timing is logical and in step with broader strategies.
As Ericsson’s Peter Linder, head of 5G marketing in North America, put it, “When we accelerated the standard and said we can do 5G at the end of 2018 rather than the end of 2020, we did not have the ability then to do both core and radio at the same time. We said, ‘Let’s focus on doing all the radio stuff first in a way that it’s as easy as we can possibly make it to connect into an existing EPC that’s upgraded with 5G capabilities.”
Speaking on Arden Media’s podcast Will 5G Change the World?, Oracle’s John Lenns, vice president of product management, sized up the standalone transition based on three types of operators: early adopters, fast followers and the mass market. With early adopters, “You’ll see some standalone architecture networks going live this calendar year.” The fast followers are “putting out requests for information to prepare themselves for issuing RFPs, and the mass market is still further out into the future.”
As far as what considerations are top of mind as operators strategize and invest in standalone 5G, Lenns highlighted security and rapid security responsiveness and cost efficiencies both capital and operating. “From a CAPEX perspective, they are looking for an efficient transition through virtualization to cloud-native. They don’t want to pay twice. From an opex perspective, they are recognizing that assembling this 5G solution…is a challenge. It’s not easy…The CSPs are looking for solutions that make that opex journey less expensive. How that manifests itself is they are looking for a solution that offers them efficiencies of deployment, more automation, more embedded test tools, more self-healing behaviour.”
“The biggest thing that will have an impact on the total costs is the automation. You have to automate as much as you possibly can.” Peter Linder, Head of 5G Marketing, Ericsson North America
Just like moving from 3G to 4G or from 4G to 5G, the shift from a non-standalone architecture to a standalone architecture is a gradual and phased movement informed by the mix of assets a particular operator has and strategic service offering priorities. Using standalone to solve for coverage is very different than using standalone and other technological capabilities to enable a smart manufacturing facility. During these phased transitions, operators will use a mix of virtualized network functions and containerized network functions running in a cloud-native core.
The co-mingling of EPC and cloud-native architectures led Ericsson to develop its dual-mode 5G core. “The difference between EPC and 5G core is essentially an architectural difference and how you operate and execute around that,” Linder said. “When we looked at all the different migration options… we came to the conclusion that the only way you could secure a smooth evolution for service providers is to combine EPC and 5G core. The dual-mode is essentially about giving the option of doing either EPC or 5G core or EPC and 5G core combined.” In that combined scenario, “You can cut and freeze the investment in the current physical and virtualized platforms. Over time you can start phasing out both physical and virtualized EPC and have everything supported by the 5G core.” Recall Lenns’ comment about not wanting to pay twice.
Linder continued: “The move from virtualized to cloud-native eliminates integration steps. People went through so much pain depending on which virtualization [solutions] they used on which hardware. Right now, moving toward cloud-native, that takes away a lot of that cost.” Another key factor he identified relates to opex. With standalone, “The biggest thing that will have an impact on the total costs is the automation. You have to automate as much as you possibly can.”
In an August 25 announcement, Verizon gave a good look at this evolutionary process from non-standalone to standalone 5G, the role of the vEPC and how it relates to network slicing and edge computing, both of which we explore further in this report. Verizon described its latest as an “end-to-end fully virtualized 5G data session,” and called it a “technology milestone [that] provides the foundation for Verizon to rapidly respond to customers’ varied latency and computing needs by providing the foundation for wide-scale mobile edge computing and network slicing.” While Verizon is planning a phased move to standalone as early as this year, this particular data session used the operators vEPC and non-standalone 5G network.
Verizon concurrently called out its RAN virtualization efforts and looked ahead to using general-purpose hardware “Instead of adding or upgrading single-purpose hardware, the move to a cloud-native, container-based virtualized architecture with standardized interfaces leads to greater flexibility, faster delivery of services, greater scalability, and improved cost efficiency in networks.”
SVP of Technology and Planning Adam Koeppe said in a statement, “Virtualizing the entire network from the core to the edge has been a massive, multi-year redesign effort of our network architecture that simplifies and modernizes our entire network.” This demonstration used vRAN equipment provided by Samsung Networks and used Intel FlexRAN software reference architecture, Xeon processor and FPGA acceleration card.
“Massive-scale IoT solutions, more robust consumer devices and solutions, AR/VR, remote healthcare, autonomous robotics in manufacturing environments, and ubiquitous smart city solutions are only some of the ways we will be able to deliver the promise of the digital world. Advancements in virtualization technology are critical steps towards that realization,” Koeppe said.
In this move from non-standalone to standalone and interworking of cloud-native and legacy systems, Rohde & Schwarz Technology Manager Andreas Roessler cautioned that there’s always a “pro and con. SA allows [operators] to implement an end-to-end service-based architecture… but one may watch out that this does not affect any loss of connectivity due to legacy technology not being supported. Meticulous network deployment is the key, as always.”
Roessler also gave an important description of the change in the access procedure when shifting from non-standalone to standalone 5G. With SA, “The UE needs to synchronize autonomously to the 5G carrier frequency and to acquire the essential system information broadcasted within SIB Type 1. Once the UE has that, it knows all the required parameters to perform the access procedure and connect with the network. The scanning process for SSB blocks is the same except the UE does not have any prior information like in NSA mode…The prioritization of SSBs is the same as in NSA mode. For the detected SSBs, the UE would measure signal quality (RSRP, RSRQ, SINR), which needs to be above a network-defined threshold. For all the detected and measured SSBs that are above the threshold, the UE will randomly but with equal probability select one SSB, and use the associated time-frequency resource in uplink direction to perform the access procedure.”
Another important theme associated with the move to not just standalone 5G but fully virtualized networks complete with edge computing capabilities is that telecom networks are becoming more like IT networks – software-defined networking and a shift toward commodity hardware is familiar to companies like HPE which sees significant opportunity in 5G.
Domenico Convertino, vice president of product management for HPE’s Communication and Media Solutions business unit, told us, “When we started looking at 5G…we were coming from a presence in the mobile core–2G, 3G, 4G–that was pretty much subscriber data management. Looking at the way 3GPP was defining the [5G NR] standard at that time, we thought that this was going to be a huge opportunity for a company like HPE. What the telcos are trying to adopt now is a transformation to cloud-native that enterprise IT started many years ago.”
He continued: “We tried to take a position, as a company, first to provide the right infrastructure for 5G because 5G is coming with different performance and scalability requirements. The second thing is to look at the access network of the mobile operator of the future– more and more convergent with an edge cloud. And from a pure software point of view, the idea was to help telcos adopt all the best practices of IT and the simplification cloud brought to IT, all those best practices that, at the end of the day, can dramatically reduce the cost of ownership.”
From a product side, HPE in March announced its Core Stack which the company describes as including “stateless containerized network functions…a shared data environment… a common platform as a service (PaaS) architecture, end-to-end management and orchestration (MANO), and automation framework, all pre-integrated on carrier-grade infrastructure as a service.” The company noted current emphasis on 5G RAN investments but said the “true value” of 5G emerges when a new core is introduced; “this enables holistic management, data sharing, and slicing into virtual 5G networks with dedicated usage and characteristics.”
HPE’s VP and GM of Communications and Media Solutions Phil Mottram tied 5G core adoption to new service-based revenue opportunities. “Investing in a new 5G network before the revenue streams are there is a financial and technical challenge for many carriers, but… telcos can start deployments today and pay for the infrastructure as their revenue grows.”
Helping telcos monetize 5G is a primary focus for VoltDB. The company sees properly handling growing data volumes and transactional speeds as key to turning 5G investments into service revenues; as such, VoltDB is helping operators make decisions in the sub-10 millisecond range.
Chief Product Officer Dheeraj Remella explained, “Data is not a database problem anymore. Data is a data problem. What is the value you’re missing with the existing choices that you can get if you think differently?” In the context of standalone 5G and a containerized, microservices-based approach, “You’re not only storing data and asking for data but rather you are using this platform to signal the next component in your service mesh to start doing its job as soon as it’s required. You have to have data storage and data stream processing capabilities to have a cohesive service flow.”
Operator focus: T-Mobile
In Early August, T-Mobile claimed a world’s first with the launch of a nationwide standalone 5G network that uses its 600 MHz spectrum. T-Mo initially launched its 600 MHz 5G network last year and reached nationwide coverage–200 million people covered–using the non-standalone architecture. With the shift to standalone, the operator saw a coverage expansion of 30% to 1.3 million square miles, upping population coverage to 250 million, and a 40% reduction in latency. Before we look ahead, let’s look back at some milestones in T-Mobile’s journey to standalone. In August 2019, T-Mobile completed a standalone 5G over the air data session using the multi-vendor kit in a Bellevue, Washington lab. Vendor support came from Ericsson, Nokia, Cisco and MediaTek.
T-Mo announced another round of standalone activities in May of this year. Working with Ericsson, the companies completed a standalone 5G data session between commercial modems on a production network. They also completed a low-band standalone 5G voice call with a mechanism to fallback to Voice over LTE (VoLTE), along with low-band Voice over New Radio (VoNR) and Video over New Radio (ViNR) calls.
With the standalone 5G network up and running, T-Mobile Vice President of Radio Network Technology and Strategy Karri Kuoppamaki told RCR Wireless News, “It’s a huge step forward in our evolution and our plan to bring 5G for all–to everyone everywhere.”
T-Mobile is following a 5G spectral strategy it often compares to a layer cake. The low-band 600 MHz network provides wide-area coverage, 2.5 GHz spectrum acquired from Sprint brings a balance of coverage and capacity, and millimetre-wave deployment is reserved for urban cores and other dense user environments.
“The benefit of SA is that it sort of breaks the dependency on mid-band spectrum which is sort of the anchor for 5G in non-standalone mode,” Kuoppamaki said. “This then allows us to bring 5G on low-band to areas that didn’t have 5G.” But the transition is about more than just coverage expansion, he said. “The last [benefit], which I think is probably one of the most important benefits of this as well, is that it’s really sort of a key to our 5G future and many of these advanced features that talked about in 5G.”
Asked about VoNR, Kuoppamaki said it’s not supported today but, “We’re working very hard to introduce Voice over NR.” Discussing how traffic is managed in areas where standalone 600 MHz 5G and non-standalone 2.5 GHz 5G are both available, he said, “Non-standalone and standalone are not mutually exclusive,” noting the ability to transition between the networks based on application demand from the UE. “This is an ever-changing scenario,” he said.
“What drives us is obviously the best customer experience and the best speed experience. I’d say that there are a couple of different cornerstones to our strategy. One is to push 5G evolution forward very, very aggressively. We’re never going to be happy with where it’s at any point in time. The second piece is to deploy the spectrum assets we have more broadly. The third one is just in general to improve the network and its coverage across the board. I think those are the types of things that are pushing us over time.”
“It’s a huge step forward in our evolution and our plan to bring 5G for all – to everyone, everywhere.” – Karri Kuoppamaki, Vice President of Radio Network Technology and Strategy, T-Mobile
Operator focus: Rakuten Mobile
Rakuten Mobile, the mobile operator subsidiary of Japanese e-commerce giant Rakuten, launched LTE services this year on top of a fully-virtualized, greenfield network composed of 330 “far edge” sites connected to 58 regional data centres hosting vRAN workloads, and three central data centres that primarily host control plane workloads. The operator currently uses a virtualized evolved packet core built following a Control and User Plane Separation (CUPS) architecture for LTE services.
Speaking during Light Reading’s 5G Networking Digital Symposium in June, Rakuten Mobile EVP and CTO Tareq Amin discussed the company’s roadmap for evolving its core network to non-standalone and standalone architecture. On CUPs, “We felt this was mandatory and necessary if you wanted to offer local breakout and true edge applications. We really pushed very hard to enable this and enable it at scale.”
With its vEPC, Rakuten Mobile has used a microservices-based architecture wherein software is decomposed into loosely coupled bits that can be rapidly rearranged into various network functions. “In our network,” Amin said, “because of the microservices architecture that we have implemented, we could really have [an] infinite number of UPF and control plane functions. Whatever happens at any instance of time in data centre one, we could in real time be able to carry the session in data centre two or three and be able to manage this traffic.” This ability to automatically move a workload “started to point to possibilities and ideas” about what a cloud-native architecture can enable.
While Amin said the 198 unique virtual network functions Rakuten Mobile has deployed as virtual machines running on OpenStack is “amazing” compared to a proprietary implementation, “There is a lot of things that are missing–quite a bit actually. As elegant as this VM architecture that we have done is, we are not completely satisfied. We need to get to a state in which we are able to truly, truly have elasticity… You never worry about capacity anymore.” Which brings us to the coming transition of the core network to a cloud-native architecture.
Looking ahead to the activation of non-standalone 5G and then the transition to standalone, as well as the future trajectory for the LTE network’s vEPC, Amin said the mission is to deploy all of it on the company’s own cloud platform. Rakuten Mobile is working with compatriot firm NEC on what Amin described as an “open core.” That collaboration considers joint development of a container-based standalone core using source code developed by NEC. The two firms are also collaborating on the manufacture of 5G radio units.
Amin laid out his thinking on using the LTE vEPC to support a non-standalone launch, then discussed the next step. “The most challenging thing in the cloud-native 5G core, in my opinion, is the completion of a highly-scalable, high-throughput UPF. I think the control plane functions are relatively straightforward. We want to achieve a very good throughput on our UPF containerized architecture. We’re spending considerable time with NEC on the development of that feature. I don’t think NSA is an exciting thing whatsoever. It just gets us out there with higher bandwidth and higher speed for the end-user. This is not where we want to be. When we launch our 5G core, for a period of time we will run them in parallel. But 5G core, once built with all containerized functions and components, will collapse all the 5G functions” into a single, converged cloud-native core.
“In the 5G core era, everything that we do must start with cloud-native. It has to have the personality and the architecture of microservices.” – Tareq Amin, Executive Vice President and Chief Technology Officer, Rakuten Mobile
Operator focus: Vodafone UK
Vodafone U.K. sees standalone 5G as a key enabler of advanced use cases like autonomous vehicles, smart manufacturing, remote surgery and the “internet of senses,” according to materials published by the operator. In pursuit of that future, the operator in July deployed a standalone 5G network at Coventry University that will be initially used to enable virtual reality-based training for “student nurses and allied health professionals,” the company said. The network uses equipment from Ericsson, MediaTek, OPPO and Qualcomm.
Coventry University’s Vice-Chancellor John Latham said the standalone 5G network “will help us continue to change and enhance the way students learn” and added that the institution’s goal is “creating a 5G campus…We will soon be able to reveal how we will use this technology to maximize the potential of virtual reality teaching for our Health and Life Sciences students.”
Vodafone U.K.’s Chief Technology Officer Scott Petty said the current focus of 5G is “increased speeds… but it’s only the tip of the iceberg of what 5G can do. With this new live network, we’re demonstrating the future potential of 5G and how it will be so valuable to the U.K. economy… From here, we will really start to see 5G make a difference to the way organizations think about being connected, and what’s possible with connectivity in the future.”
Among the chief benefits standalone 5G enables, Vodafone U.K. calls out network slicing (more on this later). The operator also notes that the distribution of computing power closer to where data is generated is key to fully realizing the latency reductions made possible by standalone.
Vodafone U.K. has offered commercial 5G service since July 2019 and has availability in dozens of cities as well as elsewhere in its multi-national footprint, including Ireland, Italy, Germany and Spain.
“From here, we will really start to see 5G make a difference to the way organisations think about being connected, and what’s possible with connectivity in the future.” – Scott Petty, Chief Technology Officer, Vodafone U.K.
Leveraging latency requires decentralized compute
As we’ve established, latency reduction is a big selling point for standalone 5G networks. Single millisecond latency, combined with ultra high capacity and speeds, opens up real-time use cases– things involving autonomous assets, precision robotics and integration of augmented and virtual reality into business processes. But all of these applications depend on real-time data creation, transport, analysis and the action initiated by that process. Regardless of the latency on an airlink, if that data has to be transported to a central processing facility, it’s a wash. This is the argument for decentralizing data centre functionality to edge compute nodes.
Speaking of edge and core, Linder said, “Those two are kind of yin and yang in terms of functionality. We expect them to be at the same location. If you bring out edge compute to say 20 locations in Dallas, and then pipe all the traffic back from those edge computing sites to a cloud core node in Austin, then back out to the subscribers, you’ve lost all advantage. The core network functionality has to be close or closer to the subscribers or you lose all the advantages.”
Remella proposed that, in a decentralized network capable of supporting huge bi-directional data flows, a mobile edge compute node “becomes a representation of the core. The MEC node can actually house the core. If the entire point of 5G is fatter pipes and lower latency, you can’t travel 1,500 miles [to a centralized data centre] even if it’s over fibre and retain the low latency.”
The data velocity involved in applications enabled by standalone 5G, as well as the autonomous manner of the delivery, creates the need for a multi-faceted approach to data handling. Remella said VoltDB things about a fast cycle and a slow cycle.
“When you look at a 1-millisecond latency network, these things are happening very fast,” he explained. “When an event happens, the next 10 milliseconds are really, really important for you and you need to do a lot of comprehensive things in that window to be able to monitor or to be able to provide quality or SLA assurance or detect and mitigate a threat.”
He described the slow cycle as driving “automated intelligence. As network events are happening, you need to do the fast cycle and siphon data into the slow cycle for machine learning. The fast and slow cycles need to play in tandem. It’s not a client/server modality. We’re seeing the confluence of a database platform and streaming platform coming together to solve one complex problem.”
“The MEC node can actually house the core. If the entire point of 5G is fatter pipes and lower latency, you can’t travel 1,500 miles [to a centralized data center]…and retain the low latency.” – Dheeraj Remella, Chief Product Officer, VoltDB
New slices, new services
The ability to deliver a network slice is the long goal of the transition to standalone 5G. At a high-level, a network slice is an end-to-end logical partition of a network that provides specific levels of service in an autonomous fashion. This can take numerous forms but the high-end vision is an operator providing an enterprise with its own slice capable of flexibly delivering everything from low-power sensor connectivity to real-time data streaming and analysis. The enterprise gets everything it needs in terms of connectivity and the operator provides a differentiated service in a manner that optimizes the use of network and spectral resources. But, like most things in telecom, it won’t happen overnight.
“It’s something that’s going to mature in steps,” Linder said, analogizing the process to painting traffic demarcations on a road. “Perhaps the first step is to get the white paint on the sides so you get the traffic on the road. The next step is a little like putting the stripes in the middle and putting in the lanes.
In an early form, Linder considered one slice comprising the public network and a second slice supporting private-public safety communications. “Then, as you move further, we can discuss here should the slices be based…on the use case or use case categories?” Slices for fixed wireless access, mobile broadband and the internet of things, for example. “When it’s very small, perhaps getting to two [slices] is a step but when you go beyond two to four or six, what is the logical step for logical compartmentalization? I see it as something that’s going to grow and develop and gradually get refined.”
Roessler sees network slicing as something an operator will offer when there’s a need to “serve significantly different QoS. Network slices should be selected based on the UE type and its requirements. For example, a local utility provider deploying smart meters has specific QoS requirements that network slicing can address also using a particular frequency band…for extended coverage and uses advanced Rel-16 features like 2-step RACH to allow quick access to the network transmitting small data packets. Applying network slicing is just one step to get the full benefits…Also, the correct features on the air interface and the infrastructure need to be implemented by chipset and terminal vendors as well as infrastructure providers.”
When considering network slicing, Roessler said considerations must extend beyond the core. It’s “drive-by software-defined radio trend as in 5G core network elements are functions…that are independent of hardware. Activating network slicing in the core is just one step. The supporting features on the air interface, i.e., bandwidth parts, mixed numerologies, etc. have to be there as well; or a flexible infrastructure with following the trend of mobile edge computing, e.g. for lower latency only a network slice is not sufficient. We have to shorten the distance between client and server to become faster.”
He also pointed out that increasingly autonomous networks give way to increasingly autonomous test and measurement practices, including network optimization, quality benchmarking and service quality monitoring. “You can install probes in a fleet of Uber cars, waste trucks, or busses that collect network data, evaluate the QoE of applications, and report the results to a central entity. This is also called ‘autonomous benchmarking’ since it is unsupervised and not linked to a drive/walk test campaign. Operators can quickly get details in real-time (and offline for post-processing in addition) about the quality that subscribers perceive in their networks.”
Remella said that for industrial and enterprise IoT-type implementation, “Standalone 5G is really key. I think that’s where standalone 5G is really going to start shining. When you look at today’s implementations, it’s a combination of 4G core…plus your [5G] hardware radio infrastructure investment. When you put these two things together, you still aren’t tapping into 5G specifically.”
The need for agile delivery of network slices is necessary for vertical digital transformation “because you have rapid data generation and consumption.
Also in this process, you have to bring more intelligence to the edge to make a real-time control loop for things like process automation and digital twins. This is where your 5G is going to really accelerate revenue.”
So what does it look like when all these pieces–5G connectivity and core alongside edge compute–are put together and applied to a use case that creates business value for a user and revenue potential for a service provider?
“Activating network slicing in the core is just one step. The supporting features on the air interface, i.e., bandwidth parts, mixed numerologies, etc. have to be there as well.” – Andreas Roessler, Technology Manager, Rohde & Schwarz
The transition from 4G to non-standalone 5G, although accelerated from a standard and deployment perspective, was and still is gradual. The same will be true for the transition from non-standalone to standalone 5G. And, in addition to investments in a cloud-native core, the full benefits of standalone 5G also require concurrent investment in virtualization and cloudification beyond the core out to radio sites and edge computing nodes. How operators approach this confluence of technologies and the management of them will be informed by assets on hand as well as strategic market priorities. But for 5G to rise up and meet the goal of enabling the broad digital transformation of enterprise and industry, “Everything is related to the 5G core,” Convertino said.