When you’re looking at your mobile network or home broadband, you know it’s important to check data speeds for app downloads and video uploads. What you might not know is that a more obscure speed measurement, called latency, is also crucial. Latency is a measure of responsiveness, with low latency a foundational part of‘ new possibilities, such as online gaming, that are technologically difficult on earlier mobile networks.
Latency is getting more attention in the 5G era, so now’s a great time for you to get up to speed, too. Rev up your knowledge with this FAQ.
What is latency?
Latency measures the time it takes for your phone to send a message and get a response. Shorter latency enables quick response interactions, which is why gaming is such an important example. Nobody wants to turn a virtual steering wheel or pull a virtual trigger then see nothing happen till after an awkward delay that could end the game.
Some people define latency as the time it takes for your phone to get a response from the cell tower at the other side of its radio link. A more useful measurement also factors in the time needed for your data to travel across the mobile network to its destination on the internet and back. The latter makes for higher latency numbers, but also more realistic ones, since that’s how we actually use mobile networks. The apps our phones interact with are on a server somewhere, not in a cell tower.
Peter Linder, head of 5G marketing at network equipment maker Ericsson, likened the magnitude of low latency’s changes to Amazon’s transformation over the last two decades. Amazon started with one book warehouse, but the company was able to expand what it sold as it brought fulfillment centers closer to customers and sped delivery all the way down to same-day shipping. That means you can use Amazon to get vastly more things. Similarly, low latency makes a network more adaptable.
What does low latency enable?
In principle, low latency 5G networks open lots of new possibilities for services that demand nearly instant response time. That includes telemedicine, augmented reality headsets and communications to let autonomous vehicles talk to one another and link into efficient platoons.
In practice, those applications are distant possibilities at best. “Nobody is going to do robotic surgery. Autonomous cars will never be controlled by 5G,” said Technalysis Research analyst Bob O’Donnell. A local surgeon would be less expensive than high-end robotic equipment operated by a remote surgeon, and cars need to communicate directly since networks are slower and far from ubiquitous.
But some low latency tasks aren’t just sci-fi. Multiplayer games running on your phone need a fast response time for you to keep up with other gamers. Games streaming over a cloud computing service like Nvidia GeForce Now or Google Stadia also need a high-speed link. And bear in mind that 5G is used for “fixed wireless” broadband links to homes, not just mobile network links to phones.
Even web surfing benefits from low latency, since so many websites are made of resources fetched from dozens of disparate servers.
How long is latency on mobile networks?
Often faster than an eye-blink.
Third-generation mobile networks — 3G to you and me — had latencies in the hundreds of milliseconds, which is to say an appreciable fraction of a second. 4G networks started with latencies of about 100ms and now are down to a range of about 30ms to 70ms, said Linder. That’s getting closer to the theoretical 4G latency of just 10ms.
With 5G networks, “We have seen networks creeping into single digits,” with good networks somewhere between 5ms and 20ms, Linder said.
But that’s just today’s latency. The ultimate goal for 5G, set by an industry group called the 3GPP (3rd Generation Partnership Project) hopes 5G network improvements ultimately can push latency all the way down to 1ms.
How do I find my network’s latency?
I like the Cloudflare speed test site best, though, because it also shows jitter, which is essentially the variation in latency over several tests. Lower jitter is good because it means your network is more reliable.
Just remember these measures of latency are useful but simplistic. They typically are just a basic connection request to a server, called a ping (the term derives from the sonar pings you hear on submarines). Real-world communications are more complex than just a ping, with more data and processing needed for gaming, video chat or other apps. Also, the server receiving the ping might be closer on the network than the real-world ones you’re reaching with your phone.
When will 1ms latency be possible?
Don’t hold your breath.
“I would say 10 years, if we’re lucky, for 1 millisecond for 5G,” said Mike Eddy, vice president of corporate development at radio technology company Resonant. “Latency is one of the most difficult and complex topics. People think there’s a silver bullet, and there isn’t.”
What does a few milliseconds of latency mean in the real world?
Networks have to respond fast if they want to keep up with the human nervous system. If data piped to augmented reality headsets doesn’t arrive fast, people get headaches as what they see lags their head motion. If a device also has tactile feedback, that has to synchronize with what we see within 7ms, said Jane Rygaard, Nokia’s leader for dedicated wireless networks.
That human-machine link could be important for an engineer performing repairs with an AR headset that overlays information about the job, or for a driver following AR directions to a container for pickup at a crowded port.
Indeed, industrial uses are a big deal as companies use their own private 5G networks to oversee equipment at ports, mines and airports. Equipment maker Nokia has 290 such customers already.
Vehicle-to-vehicle communications — the V2V, part of the— also needs a fast response time for some radically different driving technology. For example, latency needs to be just 1.5ms for vehicles to coordinate themselves into a platoon, said Jim Misener, leader of Qualcomm’s V2X work.
Why do we have to wait so long for low-latency 5G?
Because there’s a lot more equipment to upgrade besides just the radio link between your 5G phone and the local cell tower.
That includes upgrades to the cell tower’s connection to a carrier’s “core network,” and that network’s connection to the internet.
Today’s cell towers often use technology called a nonstandalone (NSA) network, a hybridization of 4G and 5G. That enables 5G’s fast data transfer speeds over a phone’s radio link, but still has 4G limits for controlling network operations.
Upgrades to 5G standalone (SA) technology sidelines the slower 4G equipment for 5G connections. In the United States, T-Mobile began its move to standalone networks in 2020 and likes to brag about that advantage over AT&T and Verizon.
“We’re the only company in the US that has launched an SA network,” and SA is required for low latency and the experiences it makes possible, said T-Mobile Technology President Neville Ray.
So telecommunications companies have to spend a lot of money to make low latency real?
Yes, and those capital expenditures are big — an estimated $69 billion in 2021 for the US carriers’ network investments, according to GlobalData.
“Every single place where a cell tower connects to the core network has to be upgraded. That takes a long time,” O’Donnell said.
But it’s not just telco giants. Cloud computing companies including Amazon Web Services, Microsoft Azure and Google Cloud also are investing in data center equipment closer to the telco networks for faster communication links. This decentralization, typically at the level of major metropolitan areas, means faster response times for people in urban areas. “Each 100 miles of fiber optics adds 1ms to the delay,” Linder said.
New standards work also is underway. Today’s standalone technology 5G was standardized in 2019 with 3GPP’s Release 15, but Release 16 is working its way to market and Release 17 is under development for completion in 2022.
A related standard that will benefit, called Ultra Reliable Low Latency Communications (URLLC) is designed to guarantee super-responsive networks. That’s mostly useful for business functions like running factory equipment. URLLC allows telcos to upgrade with technology called network slicing that dedicates particular resources to URLLC and other specific uses.
Further gains should come from 3GPP standards arriving later this decade with 3GPP Release 18, 19 and 20, collectively called 5G Advanced, said John Smee, Qualcomm’s vice president of engineering. Lower latencies there could help cars communicate with pedestrians’ phones and let phones use AI services running on powerful cloud computing machines, he said.
CNET’s Eli Blumenthal contributed to this report.