Technalysis Research
 
Previous Columns

February 13, 2020
T-Mobile, Sprint Merger Likely to Bolster US Competitiveness for 5G

February 11, 2020
Samsung S20+ And Ultra Launch Finally Brings “Full 5G” to Market

February 3, 2020
The Top 5 Fallacies About 5G

January 9, 2020
CES Previews What to Expect from 5G in 2020

2019 Forbes Columns

















Forbes Columns
TECHnalysis Research president Bob O'Donnell writes a regular column in Forbes and those columns are posted here and archived on this site.


February 18, 2020
5G Latency Improvements Are Still Lagging

By Bob O'Donnell

In an age of instant gratification, we’ve become accustomed to not having to wait for things. From fast food to instant ordering to same-day delivery, to continuously updated software to faster interactions with just about anything with which we are connected, our distaste for waiting is apparent at many levels. (Now, whether that’s really a good thing is a whole other discussion….)

Perhaps in that light, it makes sense that one of the most eagerly awaited improvements for 5G has been important reductions in what’s technically referred as latency, but what essentially boils down to lag time or delay. As with any kind of wired or wireless connection, there is always a certain amount of delay between when an action or a request is made on one device and when a response is received back on that same device. In between, the message has to be formed on the sending device, the packets of information that make up that message sent over the network, the data received on the other end, the appropriate action taken, new data gathered based on the request, then that new data sent back over the network, and the response received and acted upon by the originally requested device.

Despite all of these steps, the amount of time it takes to do all this is incredibly short—even on today’s cellular and WiFi networks, it’s measured in tens of milliseconds. (As a point of reference, average human reaction times range from about 100 to 300 milliseconds, although our brains take about 13 milliseconds to register what our eyes see; and we can hear audio delays as small as one millisecond.) One common way to measure all of this on a wireless network is to use a tool like the SpeedTest app and notice the Ping time—which detects how long it takes to get a response from a quick message sent by your device to a connected network.

One of the promises of 5G is that it’s supposed to reduce latency times down to 1 millisecond, which, at first glance, certainly sounds impressive. As with many aspects of 5G, however, it turns out the full story isn’t quite that simple. Plus, unbeknownst to many, the latest version of 4G LTE Advanced Pro has evolved to the point that it does a pretty good job on latency as well.

To understand the latency issue completely, we need to break it down to several different elements. First, the manner in which messages are created, encoded, and sent over a network can have an impact on latency. This is when enhancements to the 5G NR spec do make a positive—but actually really small—difference. Essentially, 5G NR is more efficient doing this than 4G, but the net improvement is less than one millisecond. The frequency of the signal being used to transmit also has an impact on this speed. Therefore, the fact that 5G can use high frequency millimeter wave signals does improve things, but by less than one millisecond. In real-world terms, it takes both 4G signals and low-band (below 1 GHz) 5G signals about one millisecond to travel from your device to a cell tower. It takes mid-band 5G signals about ½ a millisecond, and mmWave signals can make this trip in about ¼ a millisecond. So, technically, yes, 5G can be faster, but not in a way that is at all perceivable by a human being.

A much bigger factor in influencing latency times has to do with what happens to the signal after it reaches the tower. This is where the different 5G network modes come into play. All current 5G networks use what’s called Non Standalone or NSA mode (see “The 4G-5G Connection” for more). What that means is that the radio connection from your smartphone to the tower is 5G, but from the tower to the rest of the “core” network is still 4G. The move to Standalone 5G networks—which replace the 4G core with a new, faster 5G core—is what’s needed to really experience the kind of latency improvements that we were promised with 5G. Unfortunately, while we might see a few experiments in 5G standalone networks this year, widespread deployments of 5G standalone networks in the US are still several years away.

In case you’re wondering, the reason 5G core networks will allow for reduced latency times is because of a technology called network slicing. As the name implies, network slicing cuts the overall network bandwidth into chunks or lanes. It also provides mechanisms to manage those lanes individually and offers certain types of capabilities on individual lanes. For latency purposes, network slicing can ensure that certain lanes have less traffic than others—think of them like the HOV lanes on a freeway—thereby ensuring that data travelling on those lanes will do it without delay—hence the reductions in latency. As with a highway, you can’t do this on every lane, and you may have to pay for the privilege of using those lanes. But at least standalone 5G networks with network slicing support will enable the option.

Even with these enhancements in place, however, it’s important to get a reality check on what public 5G networks can offer in terms of latency. While final numbers will depend on real-world deployments, my understanding in speaking with 5G network experts from companies like Qualcomm, National Instruments, AT&T, T-Mobile and others is that numbers in the 10-12 millisecond range are much more likely than 1-2 milliseconds. To be fair, it may be possible to achieve these single digit millisecond delay times on private, industrial 5G networks—indeed, that’s part of the promise of the URLLC (Ultra Reliable Low Latency Communications) spec that’s being finalized in the 3GPP Release 16 document scheduled for release later this year. However, these will be in more controlled environments and not subject to all the traffic that’s going to occur on public 5G networks. Another interesting point to note is that today’s 4G LTE Advanced Pro networks have already reached latency times in the low 20 millisecond range, so while the 5G enhancements may be nice, they won’t provide dramatic improvements to the latencies we’re already seeing today.

Latency issues can actually be tackled in several different ways. Yet another enhancement that we’re likely to see over time doesn’t really have anything to do with 5G, but rather with how cellular network infrastructures are evolving alongside 5G. Specifically, we’ve started to see big network carriers like AT&T and Verizon start to integrate more cloud computing technologies directly into their towers. Late last year, in fact, AT&T announced a major deal with Microsoft’s Azure, which was soon followed by one between Verizon and Amazon’s AWS. In both cases, these technology agreements will allow the wireless carriers to integrate and run edge-based versions of their respective cloud computing platforms within the wireless network infrastructure of the carries. In other words, instead of having to take a data request received at a cell tower and then send it to cloud computing infrastructure located somewhere else within the network, the request can be responded to at the tower and then sent directly back to the phone.

In practical terms, it’s hard to estimate exactly how much real-world improvement this will enable (especially because it would be impossible to run all the AWS or Azure workloads directly on individual towers), but there’s no doubt that as this technology becomes more widespread it should improve wireless network latencies as well.

So, yes, eventually we’ll start to see the kind of latency improvements that should make cloud-based gaming more responsive, network streaming less glitchy, and overall wireless performance better than it’s been. For a while, however, as ironic as it may sound, when it comes to 5G latency improvements, we’re just going to have to wait.

Disclosure: TECHnalysis Research is a tech industry market research and consulting firm and, like all companies in that field, works with many technology vendors as clients, some of whom may be listed in this article.

Here’s a link to the original column: https://www.forbes.com/sites/bobodonnell/2020/02/18/5g-latency-improvements-are-still-lagging/

Forbes columnist Bob O'Donnell is the president and chief analyst of TECHnalysis Research, a market research and consulting firm that provides strategic consulting and market research services to the technology industry and professional financial community.