Hacker News new | ask | show | jobs
by johnm1019 2671 days ago
I mostly agree but I want to nitpick.

One of the areas I've heard should be improving with 5G is latency. On 4G we're talking O(10ms)? which on 5G, so says the marketing, should be O(1ms). If this is true, it will make a noticeable impact on the responsiveness of web driven apps (aka almost all apps) as well as video/audio chat liveness. Was this oversold to me (highly probable)?

5 comments

Latency is an active research area and a driver for updating wireless networks. You won't see the results in the current 5G roll outs, which concentrate on eMBB, but the research is feeding into the future URLLC (Ultra-Reliable Low-Latency Communications), which is a 5G protocol aimed at machine-to-machine communications.

Beyond URLLC are protocols aimed at replacing wires in "serious" industrial control systems: things like real time control of grid switchgear and heavy machinery, so you might not see it on your phone. Latencies of the order of microseconds have been demonstrated in recent years and the stretch goal is about 200ns.

Reducing latency between the tower and your phone by 9ms won't make a big difference to the overall experience of web apps. But I think real world 4G latency is more like 40-50ms?

In addition to mobile network latency you have to add a few more things - depending on how far you are from the data centre serving the web app you're looking at ~20ms-300ms additional network latency, plus however long the server takes to process your request, regardless of 4G or 5G. Check out https://www.cloudping.info/ for some ping times to AWS datacentres around the world

FYI O(10ms) means anything from 10-99ms.
> O(10ms)? […] so says the marketing

https://i.imgur.com/Dg4Jit5.png

That's my RTT ping to 8.8.8.8 as I travelled south on CalTrain from SF last week, on T~mobile's network.

Now, this is on an old phone that is only capable of HSPA+. However, the actual performance on the whole is still nowhere near any theoretical limits, or even limits of HSPA+.

Latency is the sum of its parts. It doesn't matter if the latency from the cell tower to your phone is 1ms or 10ms if the rest of the trip takes 100ms. It's nice to keep some of the latency down, but it's not going to magically enable 1ms transatlantic communication.
100ms is generally referred to as instantaneous in regards to user experience.

For example: A 10ms video refresh rate would be 100fps.. the average movie is 24fps.

That depends on what the user did, in video games a 100ms delay would be terrible, however in that situation the user is inputing highly timed inputs and is focused on the output.

But in general for web browsing, 100ms would be unnoticeable yeah.