Plot PRBs and MCS with it.
If PRBs and MCS is ramping slowly then its eNBs algorithm.
You can tweak some bits there but I’d argue its not much beneficial.
Ok sure, thanks for the suggestion.
Uplink throughput mainly depends on BSR and PHR, please plot those 2 too.
And of course on scheduling grants for Uplink.
Sure, but currently the problem is reported only in DL.
Will do that if the problem is reported in UL too.
From DL tput make sure the buffer an eNodeB is always full.
You have RAN online traces for this.
This is how the BO looks like in DU (eNB).
I agree there is not much BO initially.
BO = Buffer Occupancy.
So you see? Not enough data is coming from higher layers.
So on S1-U it does not come enough data for this UE.
Go check with core engineers from SGW.
UE is starviing from data from higher layers.
Sure, thanks.
Sometime in transmission they have all kind of flow-controls.
This could also be an issue.
Yes there is PDCP flow control mechanism between CU and DU but the problem persists even without flow control mechanism.
1 Like
There are features to improve the slow start.
In Huawei you have 3 functions under (TCP PROXY)
Slow start acceleration.
Also under video experience optimization you have initial acceleration functions.
Air interface acceleration guarantee.
Enhanced TCP link setup acceleration.
1 Like
Hi Pirtam
Did you manage to solve the issue ?
Hi @Mohamed_Abdelbasset ,
Yes, there were packet drops in CU side during the initial phase. CU Team fixed it.
1 Like