Improve VoLTE Latency

Hello Experts.
I am looking for some suggestion to improve VoLTE Latency.
I hope many of experts did some trial before to improve it.
Well so far what we here is below:

  1. “DRXMode” specifically for VoLTE User in SEO region and we observed excellent improvement in volte latency
  2. Also I did the trial on pre-scheduling for Volte user , but doesn’t improve anything on volte part , LTE side good improvement .
  3. We also enable new feature “downlink user plane reduction” in Ericsson but LTE latency improved but no improvement in VoLTE.

It’s Ericsson system.

Any idea or suggestion to improve further?
Looking forwards for your reply.

Hi, for VoLTE part you can check tdiscard timer linked with pdcp profile used for VoLTE qci.
Try to reduce it.
Also drx profile for VoLTE not recommended as it leads to call drop increase for VoLTE can you check pre and post call drop rate after turning on drx.

If you know for Ericsson System, then please help to specific timer.
I missed above “DRX” we already off for VoLTE user.

Try to optmise the scheduler config, timer based on TTI bundling too.
Also VoLTE codec rate change at edges also can be enabled in you have 20.Q1 above.

All are activated already.
It doesn’t work until now.
We are at 21.Q2.

Core Network Cisco, E, H supports post their rel 15 upgrades.

Pdcp discard timer for qci profile can be optimized for latency improvement.

I know for Nokia it’s tdiscard we use.

Understand, but more over its related with UE support.
We try all latest UE handset but codec rate change doesn’t configure.
I think in Ericsson already have TR.
Feature Called VoLTE rate recommendation.

Let me check.

Hi Experts.
What should be Latency in VoLTE (ms) in KPI?
Is less than 10 msec ok?
Or it should be less than that.

It is impossible to be 10 msec.

60 ms seen. But vary from vendor to vendor.

Depends on what you mean by Latency.

I need to check formula. I too believe it’s not possible but, I have KPI sheet, after getting OSS i will see formula (as I too think it’s impossible).

Need to confirm formula as this is what I saw in KPI sheet.
But normally in QCI9 its up to 100 msec I believe.
VoLTE should be less as its real time.

But I think 25 to 30 msec can be possible.

In our case, Latency should be <= 40 ms for SCFT report to pass.

This is KPI.
So may be ping value.
As latency 40 ms in Drive Test.
But latency is getting response from server and ping is just pinging to server.
Are they both same? I am confused here… :frowning:

Latency is response needed from server but is ping just doing one way in short pinging to server.

Ping latency provides round trip.

So both are same.

For more realistic latency with ping command, change the size of the packet to same bytes as network MTU.
Some ping softwares even allow to specify protocol i.e. UDP or TCP.

So network MTU packet size defines what this set normally.
So here latency in KPI seems less weird.
So it means QCI9 has space for throughput.
Normally when I was in Huawei in ping test OSS I used to keep.1500.
Sctplnk and iprt both to see if there is ping issues seen.