The 5G era has begun, and while next-generation network architectures, applications and usage trends could take years to evolve, the builders and providers of the transport infrastructure supporting the mobile industry already are preparing for the day when the most ambitious 5G hopes finally become reality.
Most 5G connectivity offered now leverages 4G network cores, per the 3GPP Release 15 standard enabling 5G New Radio to be anchored to 4G network core infrastructure. Because there aren’t many 5G users thus far, mobile carriers haven’t had to make many adjustments yet in how that traffic is transported end-to-end. But as 5G subscriber numbers and usage grows, and the mobile ecosystem and its customers start to dream bigger about what 5G is capable of, transport infrastructure will play an increasingly critical role in helping the mobile industry adapt, according to Mark Gilmour, vice president of mobile connectivity solutions for Colt Technology Services.
(Gilmour will be speaking Friday, March 27 on the FierceWireless 5G Blitz Week panel “5G Transport: Challenges and Opportunities.”).
Leveraging 4G cores, “enabled faster 5G rollouts, but also opened the air interface -- the bit between the end user and the network -- which in turn requires more capacity in the backhaul, aggregation and core network because you are opening the floodgates a bit, and bringing more capacity onto the network,” Gilmour said, although he admitted that enhanced mobile broadband has had a limited impact on transport pipes thus far.
Brian Lavallee, senior director of solutions marketing at Ciena, said initial 5G rollouts haven’t required extensive swap-outs of switching and routing equipment just yet, either, but that day will come. “We actually have a lot of deployments around the world where we are using products that were designed for 4G to increase the capacity on the backhaul for early 5G,” he said, but added, “What 5G eventually brings along with it is low latency anywhere from 1-5 milliseconds, end-to-end which is a huge decrease. What that entails is everything in that service path has to be very low latency in terms of switching and routing and so on.”
Ciena recently launched a new family of transport routers with targeted support for advanced capabilities inherent to the 3GPP Releases 16 and 17, such as that low latency, as well as the ability to carve logical slices from the physical 5G network and dedicate them to specific uses.
Those capabilities aren’t common yet, but as 5G launches increase and coverage expands, one of the first things that any 5G operator needs to do is work with transport network partners to make sure that backhaul capacity is upgraded to handle the eventual growth in and evolving complexity of 5G traffic and applications.
“The currency of the backhaul has moved from 1 Gig and below to a 10-Gig interface for backhaul,” Gilmour noted. “That’s a pretty significant jump and the big thing companies like us have jumped into in our discussions with mobile carriers.”
Brian Daniels, senior vice president of strategic networks, Z5G at Zayo, was in tune with Gilmour’s assessment. “We're upgrading our core infrastructure. In markets where we were 1 Gig we’re now 10 Gig, with 100 Gig on cores, just to make sure we can match the upgrade needs that the carriers have,” he said.
Beyond bandwidth planning, many companies need to further rethink how their network architectures should be designed to best handle the next phases of the 5G evolution. Gilmour and Daniels both suggested the imminent needs to upgrade and adapt have driven deep strategic discussions between their companies and their mobile carrier customers.
“The very first thing we have been majorly involved in the last couple of years with operators in our major markets is how do they future-proof connectivity to the macro sites, which are the underpinning of that 5G infrastructure,” Gilmour said. “The demands of bandwidth and connectivity to those sites will keep increasing. Along with that, how does their infrastructure cope with the next decade of change? The forward-thinking operators are having those partner-led discussions now with the likes of us and others.”
One major indicator of those network changes is the evolution from distributed base station architectures toward Cloud (or Centralized) RANs (C-RANs) that centralize management of more cell sites and small cells in increasingly dense networks. More computing power also is required closer to the network edge, all of which increases the need for fiber deeper into the RAN.
Many early 5G implementations use high-frequency millimeter wave technology requiring shorter distances between transmission sites, and this along with strategies to increase urban and in-building through small cell deployment has put more focus on the importance of high-capacity front-haul and mid-haul needs. C-RANs enable the baseband signal processing to occur in the cloud, for example through a unit at the bottom of a cell tower or elsewhere nearby, separate from the remote radio head functions at the top of tower masts.
The Common Public Radio Interface (CPRI) traditionally has been used for RAN links, but the 5G evolution will require a more robust interface--enhanced CPRI (eCPRI) or open RAN (O-RAN), for example--and transport bandwidth to match.
“Operators are moving away from CPRI because it can’t handle the 5G era bandwidth needs,” Gilmour said. “For example, if you’ve got a 150 Mbps user plane, or actual data plane, today for an LTE network at 20 MHz, and you are using CPRI front-haul to carry 150 Mbps of user plane or data traffic, you actually need 2.5 Gig of CPRI because there is no compression in there,” he said. “It’s akin to taking a network box, breaking it open and transporting everything on that. It’s always on. Even if there’s no traffic, you’re always transmitting 2.5 Gig, and that’s for every antenna. Typically, on a cell site you have at least three antennas. So, for 150 Mbps of data plane in a 360-degree view, you need three-times-2.5 Gig of CPRI capacity, and it linearly scales from there.”
Not many operators have gotten that far with front-haul yet, however, Gilmour noted. “Verizon, NTT DoCoMo, SK Telecom in Korea, for example, where they inherently in the same company have a fiber network - those companies have dabbled in front-haul over a wide area network.”
As mobile operators continue 5G upgrades, and increasingly begin using eCPRI and/or O-RAN in their networks, it’s not clear which approach will dominate, but regardless, the underlying transport layer for front-haul in either case is likely to be fiber-based and Ethernet, Gilmour said. The same is true of the evolving mid-haul segment, where the centralized baseband units collecting all that front-hauled traffic also will require robust fiber transport.
“As you start to see more C-RANs or C-RAN hubs in a market, that’s where you start to see the need for 10 Gig for transport if they are not using fiber already,” Zayo’s Daniels said.
Still, some mobile operators may continue to dabble in and test fiber alternatives. Both Verizon and AT&T, for example, reportedly have been eyeing Integrated Access and Backhaul (IAB), which could use wireless links in some instances instead of fiber for backhaul. However, it’s unclear to what extent IAB could be implemented (Verizon did not respond to a request for comment), and it doesn’t look like fiber is in danger of losing its job anytime soon.
In-Building Coverage, Lower Latency and Network Slicing
Bandwidth growth with the ongoing activation and expansion of 5G is one factor in the architectural evolution of networks, but other change agents include the advent of edge computing, expected improvements in in-building coverage from denser networks, further development of ultra-low latency applications and customer-driven demand for network slicing.
This mix of factors makes for an increasingly complex network architecture picture. “You will have front-haul, mid-haul and backhaul mixed in the same geographic area based on the applications that are being used,” Gilmour said. “We have this view from all of our nice drawings of the network that these are discrete parts of the network, but the reality is that when you put it into a city, these segments overlap from a physical point of view.”
Gilmour gave an example of what he meant, relating it to a project Colt is working on to plan greater support for in-building mobile coverage. “We have 28,500 buildings connected to our network by fiber,” he said. “Could we use a front-haul or mid-haul-based O-RAN network to do a virtualized neutral host network that sits in our network edge data center, then uses our fiber footprint into all of these buildings to then connect up radio antennas that sit inside the building? So, you’ve got a mixture of front-haul and mid-haul in-building, with the same fiber infrastructure coming back. You’ve also got backhaul outside.”
Another example of the newly complex environment is becoming evident with the emergence of multi-access edge computing. Requirements for more edge compute and storage resources are forcing mobile operators to deploy those resources at cell towers, larger cell sites or in nearby data centers.
That places additional importance on fiber depth and density. How fiber could be leveraged to connect this environment is much different than how it was used to support mobile networks back in the 3G days, according to Zayo’s Daniels. Back then, metro fiber rings collected traffic from macro cell sites, and backhauled the traffic to mobile switching centers (MSCs). “There is inherently a lot of latency in that model because you’re taking these longer runs around metros, passing through a lot of optronics, and each one of those hops adds time,”he said.
Now, with 5G growth starting to ramp up and new ultra-low latency applications anticipated, Zayo and others have a better model to leverage: Denser fiber in metro networks, with more fiber in buildings and core data centers, giving them “straighter line shots” connecting multiple locations to another data center or MSC.
Transport and Network Slicing
“Then, if you want to slice the network you can do that and have the same fiber distance. The ability to cut equipment hops or distance out of the network is how you improve latency,” Daniels said, adding that use cases for ultra-low latency and network slicing capabilities already are starting to emerge in industries such as automotive, retail and shipping among others, where there is growing need to collect, analyze and act on data.
Ciena’s Lavallee suggested the advent of network slicing will serve as the ultimate test of 5G transport architectures, as each slice will be subject to guaranteed performance, which could have to with low latency thresholds, capacity networks, timing and synchronization or other key metrics.
“That means everything in the service path has to guarantee performance as the data traverses it, be it a router, a switch, some optical device with storage and compute resources.” he said. “You have to orchestrate that from end to end, which means you have to have some kind of controller for the network elements in that service path to request the right level of performance, and if they say they can support it you provision that path. Then, you have to have some kind of automated tool that can monitor that service in real time because the operators will be selling 5G services at a premium.”