HSUPA vs. UL R99 Performance

[faisaladeem]

Hi,

I have done some trials of disabling HSUPA vs. having max. HSUPA users in a cell. It turns out that UL R99 throughput and KPI performance is much better than HSUPA. R99 throughput was between 100kbps to 384kbps while HSUPA throughput was around 60kbps to 100kbps. Also HSDPA throughput, RTWP and PS DCR were much better when HSUPA was disabled. Test were done in an IBS cell in busy hour where lots of users are gathered and there is poor coverage in some areas as well. Can someone explain why HSUPA performance is worst compared to UL R99 ? Isn't HSUPA supposed to improve user experience ? (our HSUPA peak rate can only be 2.88 Mbit/s).

BR,
Faisal

[babak1349]

Hi,

I have done some trials of disabling HSUPA vs. having max. HSUPA users in a cell. It turns out that UL R99 throughput and KPI performance is much better than HSUPA. R99 throughput was between 100kbps to 384kbps while HSUPA throughput was around 60kbps to 100kbps. Also HSDPA throughput, RTWP and PS DCR were much better when HSUPA was disabled. Test were done in an IBS cell in busy hour where lots of users are gathered and there is poor coverage in some areas as well. Can someone explain why HSUPA performance is worst compared to UL R99 ? Isn't HSUPA supposed to improve user experience ? (our HSUPA peak rate can only be 2.88 Mbit/s).

BR,
Faisal

Hello,
Who is the vendor ?
how you measure the throughput ? have you don't DT and walk test or you were looking at the statistics ?
for throughput analysis looking to the stats is not recommended as it is impacted by users activities For example in E******* Technology The best RAB all the time will be allocated which is HSDPA/EUL now imagine if users doesn't upload what is going to happen ? The UL throughput or HSUPA throughput will be small while it has the HSUPA service
for R'99 of cource you have Channel switching , so if there is not data to be transferred , users will be downgraded to lower rate and if you look at the throughput of RABs which have UL 384 kbits for sure you will not see lower throughput When the users downgraded to lower UL rate due to lower throughput, for example your throughput will be 128 kbits and so on....
So it is suggested to test the throughput by using DT team not stats
Besides for HSUPA, the limiting part is added interference to own cell and also added interference for RSSI So you might consider it and remove the limitation to get better throughput
It is better to do those test in a macro site because your IBS implementation is important, are you using active element ? Do the NodeB itself has proper power control on UL ? From nodeB to Antenna you need to have linear system, if you have active element , your system might not be linear any more

BR

[faisaladeem]

Hi,

I'll try to answer your questions below:
Vendor is Huawei
Throughput tests were done through DT
IBS is based on passive DAS
Standard DL/UL power control is working on the site (no specific power control is enabled for HSUPA though)

Some questions:
How to remove the added interference/RSSI/RTWP in HSUPA case ? (some parametric approaches to limit the uplink interference are inversely proportional to UL throughput and hence counterproductive)

[babak1349]

Hi,

I'll try to answer your questions below:
Vendor is Huawei
Throughput tests were done through DT
IBS is based on passive DAS
Standard DL/UL power control is working on the site (no specific power control is enabled for HSUPA though)

Some questions:
How to remove the added interference/RSSI/RTWP in HSUPA case ? (some parametric approaches to limit the uplink interference are inversely proportional to UL throughput and hence counterproductive)
hi
I don'y know about Huawei but in E******* there are some nodeB parameters the basic one is as follow:
own cell load addition ( maximum threshold over own cell generated UL load_
ROT maximum threshold ( max threshold over RSSI)
maximum allowed throughput per user

and also others
For me it is weird as I experienced all the time higher EUL throughput please note that for HSUPA yyou can use SF32 up to 2X SF2 + 2 X SF4 so which SF is used in UL system can allocate the lower SF due to added interference

have you checked the TN limitation or even QOS limits ?
when u uploading what is UL DTX you see ?

for DL you need to check the requested physical/served/scheduled throughput among Hs-dsch DTX for HS) if you see big difference between scheduled and served with high DTX rate it shows bottleneck on TN

it is to do the test in unloaded situation

cheers

[electron]

Hi,

I have done some trials of disabling HSUPA vs. having max. HSUPA users in a cell. It turns out that UL R99 throughput and KPI performance is much better than HSUPA. R99 throughput was between 100kbps to 384kbps while HSUPA throughput was around 60kbps to 100kbps. Also HSDPA throughput, RTWP and PS DCR were much better when HSUPA was disabled. Test were done in an IBS cell in busy hour where lots of users are gathered and there is poor coverage in some areas as well. Can someone explain why HSUPA performance is worst compared to UL R99 ? Isn't HSUPA supposed to improve user experience ? (our HSUPA peak rate can only be 2.88 Mbit/s).

BR,
Faisal

Hi,

There are many reason that R99 is not better that EUL few of them i listed below which all can be verified through STS as well as DT:

In R99 there is no control over ROT but in EUL strictly it is controlled hence it will increase capacity of the cell
In R99 resource efficiency is much lower as compare to EUL
Resource holding time in R99 is much more than EUL
In EUL HARQ re-transmission is performed in RBS MAC-e but in R99 it is in RNC and in RLC layer which is not enough robust same as EUL
To improve the efficiency there will be EUL grant rotation in congested situation which in load situation very much helpful to maintain performance

Considering all above mentioned points we can come across that EUL always is better than R99. Cnsidering all above mentioned points let's mention what is the exact question we could help you .


Cheers

[faisaladeem]

Hi,

Well in theory as you said HSUPA should be better than UL R99 and this is why I am putting up this question here. Although from resource utilization point of view it is still better but I am talking specifically in terms of throughput, PS drops and RTWP/UL RSSI in a cell with 20+ HSUPA active users. When there are few users in a cell, HSUPA throughput is always better (>384 kbps) but as the users increase, the throughput degrades, drops increase and RTWP also increases to very high values. It seems it is better to disable HSUPA in such scenarios. I think Huawei basic HSUPA implementation is not very good. However there are some optional HSUPA features in Huawei promising to improve HSUPA performance but unfortunately we don't have license to try them (e.g. dynamic RoT adjustment, small target retransmission, anti-interference scheduling, IRC, turbo-IC, small-packet solution<grouped feature> etc.). Comments/suggestions are still welcome :)


37543

[babak1349]

Hi,

Well in theory as you said HSUPA should be better than UL R99 and this is why I am putting up this question here. Although from resource utilization point of view it is still better but I am talking specifically in terms of throughput, PS drops and RTWP/UL RSSI in a cell with 20+ HSUPA active users. When there are few users in a cell, HSUPA throughput is always better (>384 kbps) but as the users increase, the throughput degrades, drops increase and RTWP also increases to very high values. It seems it is better to disable HSUPA in such scenarios. I think Huawei basic HSUPA implementation is not very good. However there are some optional HSUPA features in Huawei promising to improve HSUPA performance but unfortunately we don't have license to try them (e.g. dynamic RoT adjustment, small target retransmission, anti-interference scheduling, IRC, turbo-IC, small-packet solution<grouped feature> etc.). Comments/suggestions are still welcome :)


37543
Hello
Scheduler is also apart of process of HSUPA/HSDPA allocation , how the system allocateds SF for HSUPA ? IN E******* not only demands is important but also there is mechanism to look after the added UL RSSI So there if the highest HSUPA allocation ends up the RSSI exceeds the threshold, it allocates the lower UL SF which could be between SF32 up to 2XSF2 + 2XSF4 So in the case you see low HSUPA throughput it might be due to allocation of SF32 So you need to find in Huawei what is the limitation For example the the own cell UL RSSI is limited to 3, or 5, or 10 or 55 Db over noise floor ? and remove it So how do you know when you have HSUPa which SF in UL has been allocated ? Maybe the lowest one is allocated due to higher UL RSSI and limitation imposed by parameters on scheduler ?
I suggest to investigate about Up loading and downloading separately. I meant upload and check the throughput of HSUPA
Besides HTTP/FTP are using acknowledgement mechanism , it means UE needs to acknowledge about received data and if it is limited for any reason the downloading throughput might be limited
The above 2 samples show different number of users which may impact the throughput first case you have 27 + 27 (HSDPA/HSUPA users ) in E******* always (HSUPA and HSDPA are coming together. It means it is the best RAB So totally you have 54 HSDPA users which may explain the low throughput for DL
Besides you need to look at the IUb capacity ( I meant physical capacity )
If you are using TEMS, as I mentioned you can look at :
1) HS-DSCH physical Requested throughput ( it is TEMS measuring requested throughput based on CQI)
2) HS-DSCH physical scheduled throughput which TEMS gives you based on allocated resources
3) Served Throughput which is real users throughput
4) For EUL you may look at the happy bit ratio as well and can give you an idea how users is requesting

to be honest with you normally downloading is the focus and UL rate is not get too much attention
One important thing is t onot focus only on one site it is better to see networks behaviour checking Macro sites checking cells/sites in very low load time is highly recommended


Cheer