Hi all 5G/LTE Experts.
After starting UDP traffic on 2nd UE, TCP traffic on 1st UE decline drastically.
I’m observing lot of re-transmissions and duplicated ack.
Any idea why it could happen?
Both UE share same resources, so it is normal trhoughput on 1st UE to decrease.
But retransmissions should not happen since I guess you use SU-MIMO.
Yes, I’m using single user mimo.
Are you saying drastic decrease on TCP throughput shall happen?
So its normal and its not an issue?
I am saying that 2nd UE will use some of the 1st UE resoruces so your 1st UE will not use anymore all 273 PRBs and all slots.
Got you!
So do you suggest any scenario to try in order to better understand why TCP throughput declined drastically?
I need to understand the root cause, why its happenning…
DL TCP throughput was 800 Mbps and when pumped on 2nd UE UDP throughput 1 Gbps, then DL TCP throughput on UE1 decreased to 20 Mbps!
Are you sure you are pumping enough data in Dl for both UEs?
Yes.
Issue of TCP getting decreased drastically happen only and only when I pump from Core on UDP traffic any data throughput above 500 Mbps.
But if I pump on UDP traffic 500 Mbps then no impact (no TCP throughput decrease) on TCP traffic on the second UE.
This behaviour that TCP getting decreased drastically only observed when I pump udp data traffic on the other UE above 500 Mbps.
Otherwise (when udp traffic isn’t higher than 500 Mbps) then no issue or any impact on TCP traffic and it works fine.
Got me @RFSpecialist ?
What are different TCP optimizer vendors being used in your network this is not related to above TCP discussion?
Yes, I got you. But do not know why it happens like this.
It’s only when UDP traffic is very high …500 Mbps+
It’s like UE isn’t able to send ACK. Isn’t responding.
That’s why I see retransmissions.
So do you think it is issue of UE? Hmmm …
How close are the 2 UEs?
But anyway should be scheduled different PRBs and different slots in su-mimo scenario so there should be no interference even UE are close.
On cell center both. RSRP -60.
Close. It is on cage.
I’m using/configuring 2 UEs per TTI in DL , so try / configure to use 1 UE per TTI in DL?
It should not matter as each UE gets different PRBs but give it a try, why not.
Try to use TCP uplink data and UDP uplink data?
Instead of DL traffic? And letss figure if it happens
I think since its TCP so it has TCP window. So it’s more suffers than UDP traffic. Shouldn’t compare UDP to TCP on that case?
Like I believe if increasing TCP window size then it will help or to increase SR periodic it will help.
UDP always will use all HARQ ids than TCP traffic.
Do you agree?