With new network architectures, applications, and more capacity, 5G also means multiple changes for transport, including the need to support more stringent latency and reliability requirements.
Handling complexities for 5G transport will be discussed during a panel session this Thursday during the FierceWireless 5G Blitz Week Fall, a free virtual event.
At a high-level, the biggest change for transport in the 5G era is the shift to what’s known as service- based architecture (SBA), according to Mark Gilmour, VP of Mobile Connectivity Solutions at Colt Technology Services.
Ultimately this means networks and transport are no longer a one-size-fits-all scenario.
Transport has evolved over the years, but in the past - particularly for 2G, 3G, and even in 4G - transport was relatively simple. For mobile networks, there was a cell site with a wireless connection out to the device, and then a fixed or wireless connection from the radio back to the core of the network as the backhaul transport segment.
Even though 4G brought its own transport changes, “with 5G, it is completely different,” said Kashif Hussain, director of Solutions Marketing, Network Enablement at Viavi Solutions.
For 5G, it’s no longer just about delivering voice or data, but also about people, IoT devices, and machines, among others, with considerations for Ultra-Reliable Low-Latency Communications (URLLC) supported over the same transport.
By 2026, 5G networks are expected to carry 54% of the world’s mobile data traffic, according to Ericsson’s November 2020 Mobility Report. With drastically greater capacity in 5G, transport pipes are going to be much bigger and thicker, Hussain noted, because the connection back to the core has to be as well.
But bigger, so-called “pipes,” isn’t the only difference.
Part of the changes for transport stem from disaggregation in the radio access network (RAN) and splitting up cell sites, Gilmour noted. This introduces new segments, namely midhaul.
There have been some and will still be distributed RAN and centralized RAN (C-RAN) architectures, Hussain said, but we’ll also see virtualized and open RANs, where network functions get divided and the locations of those can be separated.
While there are different ways to go about it, at a basic level, as Hussain described, it’s the concept where network functions that take place in the baseband unit are split up, and now the RAN architecture of the BBU and radio is divided into three network components – sometimes known as radio unit (RU), and now the distributed unit (DU), which is more toward the radio and centralized unit (CU) located closer to the core. Midhaul is the link between the DU and CU.
The rationale for how sites are split and disaggregated will depend on the use case, the nature of scaling, or compute needs and moving closer to the edge, for example, according to Gilmour.
“We’re breaking down the network into its component parts, and then transporting and networking that all together,” he said.
The segment from the baseband to the radio, known as fronthaul, is something that has been around for a few years, but something Gilmour said has not found its way into the wide area network space.
“[Fronthaul] has kind of been limited to the cell sites or just a kilometer or two, maybe up to 10 kilometers away from the cell site,” Gilmour said. “That will become much more prevalent in the 5G era.”
Reliability and latency
As mentioned earlier, future applications for 5G go beyond just connecting people, where reliability and latency are key and mean stringent requirements for transport.
“The real crux of 5G, and we’re only really at the very beginning of it, is about connecting people, things, and machines, and creating new use cases or new applications that can be used over cellular,” Gilmour said.
Outside of enhanced mobile broadband, URLLC for mission-critical applications and Massive Machine Type Communications (mMTC), where a massive number of devices send short packets, weren’t part of the 4G discussion.
Hussain and Gilmour each pointed to applications such as industrial use cases like factory floor automation, robotics managed by a cellular network, or connected cars, where requirements for performance, latency and reliability are much stricter.
This means the transport layer needs to cope beyond the traditional view of simply connecting a cell site back to the core and transport high bandwidth traffic for mobile broadband use cases
“It also needs to deal with low latency or lower latency applications, where you need to move the compute power closer to the end user, whether that end user is a machine or people,” Gilmour said.
Building individual networks for each application is one approach, but Gilmour says the reality of that is it’s very costly and kills the business case for many applications.
“Ideally, you want to be able to create an infrastructure that can support these different types of architecture on top,” he said. To do that, the active transport layer needs to be flexible, programmable, and intelligent to deal with the different variations. “It’s all encompassed in that idea that it’s no longer a kind of generic template for traffic.”
Synchronization, time-sensitive networking
Related to reliability and latency are certain timing and synchronization protocols that can be implemented to support the transport side.
“Your transport is evolving significantly to keep up with the application that will be delivered by 5G,” Hussain said. “Low latency and reliability are the two things coming together and the network has to be reliable, so the protocols should be reliable enough that there should be some sort of an error check where if I miss something, I can recover.”
One technology protocol that Hussain called out as likely to be seen in 5G is TSN-type (time-sensitive networking) deployments in the transport network, referring to an IEEE 802.1 standard. TSN enables Ethernet-based interfaces like eCPRI (another change for transport) to work, and since it’s time sensitive, synchronization and latency requirements based on quality of service – something also seen in the network slicing concept - can be implemented.
Synchronization is a key aspect, which Hussain emphasized is different than latency. Whereas latency is the lag time between point A and point B, synchronization can be thought of as two end devices or parts of the network that are working off the same clock to send and receive.
“If one device drifts, then there will be a problem in terms of when the packet arrives at that point,” Hussain said. So in a 5G network when you have URLLC and mMTC, ensuring synchronization to the core of the network is vital or essentially information won’t be able to be sent or received if one component is outside its allotted margin of time. “You need an extreme level of synchronization.”
Typically, in a cellular network, GPS has been used, with GPS receivers integrated into cell sites. However, with greater densification and a much more disaggregated 5G network “putting GPS at every corner and every street is going to be extremely difficult,” Hussain noted.
One thing he said will be implemented is called precision timing protocol (PTP), so that timing information can be sent to all parts of the network.
“My transport network should be able to support PTP-type synchronization techniques as well,” Hussain said, noting that’s a new burden for transport networks in the 5G era.
Fiber pushes out, and microwave sticks around
Much fatter so-called pipes are one feature of transport networks, and when it comes to 5G fiber is likely still viewed as king in terms of futureproofing and capacity for backhaul.
As a fiber provider, in the last two years Gilmour said Colt has seen and been participating in the migration of existing cell sites from wirelessly fed backhaul. That’s to beef up the transport layer as initial 5G deployments have added New Radio to existing sites to be able to deliver more data on the network.
The fiber fed segment will push to more and more cell sites as carriers densify, and be able to underpin the different use cases and services-based architecture that sits on top. A key change for the active transport layer on top of the fundamental fiber footprint, according to Gilmour, is making it programable and flexible.
“Software-defined networking in that backhaul network is critical to this,” he said. That includes dynamic routing and adding orchestration capabilities, which is something Colt introduced to its IQ network last year.
In October, Colt secured a fiber backhaul deal with mobile provider Three UK, who said the fiber would provide 20 times more backhaul capacity to sites deployed.
That said, logistically and economically, fiber won’t be viable everywhere. And other options like microwave transport won’t be going away, according to Hussain, even as fiber deployments continue to increase.
To that point, the market for microwave transmission gear grew 6% year over year in the third quarter, as the wireless backhaul market started to recover after a sharp decline in the first half of the year during COVID-19 lockdowns, according to Dell’Oro Group.
"We think there is more growth to come for this market and remain positive that demand for Microwave Transmission equipment will continue to increase,” Jimmy Yu, VP at Dell’Oro, said in a statement. “Assuming the worst of the pandemic is behind us, the economy recovers, and 5G mobile radio deployments stay on pace, we are predicting the microwave market to grow 4 percent next year to $3.1 billion."
Ericsson, one of the top six microwave vendors, predicts that by 2025 38% of backhaul connections will be microwave-based, or 62% when fiber-dense countries of China, South Korea and Japan are excluded. In its recent microwave outlook report (PDF), the vendor said a mix of fiber and microwave will be needed to support increased capacity requirements ranging from 300 Mbps to up to 20 Gbps by 2025.
According to the report, microwave will largely be used for last-mile access in urban areas, as well as aggregation links in suburban and urban
IAB as complementary
There are also emerging technologies targeting backhaul in the 5G era, like integrated access backhaul (IAB), which was standardized in 3GPP’s Release 16.
With IAB, operators can essentially use an existing 5G cell site with the airlink interface as backhaul for other nearby cell sites, meaning they can use some of the same bandwidth they’re delivering 5G services with for backhaul instead.
Verizon and Ericsson performed an IAB proof-of-concept over the summer and AT&T previously signaled plans to trial IAB. Colt’s Gilmour views IAB as a great solution in the sense that it opens up an opportunity to be able to densify quickly, but says it’s complementary and definitely not a replacement for fiber, microwave or other backhaul technologies.
As Hussain noted, it can be more costly to deploy fiber or microwave, but IAB still usually requires having a fiber fed point at one of the links that backhauls the others.
There’s also a tradeoff, Gilmour noted. “You have to sacrifice some of that very precious spectrum” for operators to be able to do their own backhaul. It could potentially be used in an environment where an operator is spectrum rich, he said, particularly with high-band millimeter wave spectrum that offers a lot of capacity but possibly not a lot of reach.
More than 120 operators have announced 5G launches, but it’s still the early days.
Transport has “to be able to deal with what’s coming, as well as what’s here today,” Gilmour said.