LTE UL Throughput higher than DL Throughput

4G LTE
,
1

Hi everyone!

I’m asking about the situation when UL throughput is higher (about 3 times) DL throughput, despite PRB utilization being about the same for DL & UL.

How can that be elaborated?

Thanks!

2

Sounds like a transport issue, Sir. Just in case, double check also for both DL and UL, MCS, TTI Utilization and BLER.

2 Likes
3

Hi.

This is usually due to misalignment of transport configurations. Highlight to your transport team and they can fix it.

2 Likes
4

Thank you so much marcengo and MR_271285 for your replies!

I’ll check on those and get back to you.

Thanks!

5

@marcengo @MR_271285 Hey everyone, that was long time ago but I wanted to get back on this. Yes! There were issues from transmission side (namely fiber links), we could verified this using FEGE packets error counters.

Thanks!

5 Likes
6

i think it is transport issue

1 Like
7

Hello @Abubakr . I think this too…
I am also having experience with high UL Throughput but low DL Throughput.

Question is: how can we convince Transmission team that this is due to misalignment in transmission?

8

Misalignment?

9

Please check QOS values at TX end.

10

Usually I get this with cells with heavy DL users.
Is PRB usage low?

11

Check the SCTPrepckt counter, if vendor is Huawei.

12

Check port configuration and its KPI port how much is it rxmax speed to confirm this query.
In Huawei it can be checked from DSP ETHERNET as how much speed it is configured and in fege port how much rx max speed you get apart from packet loss kpi and ping test at LTE end.

13

Sorry how could i export this counter from Imaster

14

Hello Experts.

Any insight for the issue?

Some of the speedtests showing Double UL speeds than DL, while other see expected high DL and expected UL speed like 230, 60 Mbps.

I believe it is directly related with the speedtest server itself but can’t explain the root cause.

If it could be related with TCP ACK packet discard or latency than both UL and DL speeds should be lower with both speedtest servers but in our case UL throughput is good for both.

1 Like
15

Can you perform the test on different hours and check if the problem is related to a specific hour (for example during rush hours)?

16

Unfortunaltely not related, both doing at rush hours.

17

Possible explanations for why you might be seeing double UL speeds than DL on some speedtests, while others show expected DL speeds:

  • The speedtest server is overloaded.
    If the speedtest server is overloaded with traffic, it may not be able to handle UL and DL traffic equally.
    This can lead to lower DL speeds than UL speeds.

  • There is a problem with the routing between your network and the speedtest server.
    If the routing between your network and the speedtest server is not optimal, it can lead to higher UL latency and lower DL speeds.

You can try using different speedtest servers. If you see the same issue on multiple speedtest servers, then the issue is likely related to your network. Try connecting to the speedtest server using a different device. If the issue only occurs on one device, then the issue is likely related to that device.

TCP ack packet discard or latency can cause both UL and DL speeds to be lower, but it is less likely to cause double UL speeds than DL.

This is because TCP is a bidirectional protocol, which means that both the sender and receiver acknowledge each other’s packets.

If TCP ack packets are being discarded or are experiencing high latency, it can cause the sender to slow down the transmission rate in order to avoid packet loss.

The most likely cause of the issue is either that the speedtest server is overloaded or that there is a problem with the routing between your network and the speedtest server.

You can try using different speedtest servers and connecting to the speedtest server using a different device to troubleshoot the issue.

18

If it’s live network, I would compare DL vs. UL resource allocation; # of HARQ ID, MCS, PRB, TBs. If more resources allocated for UL, then should expect higher Tput.

19

This is very limited information.

There are many nodes and interfaces in between end servers and users.

Best is to use PCAP traces and analyse in Wireshark.