Model Tuning Help Required

[faib]

hi all,

is there anyone who has an experience on model tuning (GSM /UMTS 2100 ) , please PM(private message) me , its urgent requirement.


Regards,
Faib

[navigateurs]

hi,
what you want to know

[ocsibacsi]

We did several model tuning for 900/1800/2100 in Europe.
We collected measurement on live network (with a scanner with BSIC decoding) and also set up test transmitters and measured the CW signal.

[optimplanner]

Hello Guys,I have a question, I can do model tuning with the nodeB transmiter?, How I do a model tuning with the measurements of the live network, I mean I have a cluster with 3 transmiter sectors and I can capture the RSSI of this 3 transmited sectors of diferent Scrambling codes, I can do a model tuning with this measurements?

[aminkhan]

I did a model tuning project for a new operator in 2007.
May be i have the whole procedure for it.I will search it and as soon as i find it will be posted.

[optimplanner]

Someone we know if I can do tuning model with a live network measurement I mean RSCP of each PSC or with RSSI of this PSC, not with external transmiters?
Thanks a lot !!!!

[zoro78]

Someone we know if I can do tuning model with a live network measurement I mean RSCP of each PSC or with RSSI of this PSC, not with external transmiters?
Thanks a lot !!!!

i tried it with live network as scanning with GSM ithink we got a good results with average error almost close to zero

[d0000h]

Someone we know if I can do tuning model with a live network measurement I mean RSCP of each PSC or with RSSI of this PSC, not with external transmiters?
Thanks a lot !!!!

The problem with UMTS is as soon as you drive from the "test" site into the next site, the signal quickly becomes interfered as there is no SHO happening while you are measuring the CPICH. Ideally for an urban model you would want to measure down to say -100 or -105 and you are not going to be able to do this in most markets and still be able to decode the PSC.
The plus side is you have LOTS of transmitters to use/choose from so you can easily drive many of them to tune the model. So you should be able to get a very accurate model, but only for a somewhat limited range of signal levels because of the limitation noted above. Which still might be good enough for what you need the model for.

-Deuce-

[optimplanner]

For this I use PC-Tel Scanner, this scanner measurement level of -120 dbm in fast scanning mode, this scanner don't realize SHO, this monitoring all PSC.
I try to use model tuning with PSC, for me is better because I can measurements a lot of signal for diferents trnasmitter.


The problem with UMTS is as soon as you drive from the "test" site into the next site, the signal quickly becomes interfered as there is no SHO happening while you are measuring the CPICH. Ideally for an urban model you would want to measure down to say -100 or -105 and you are not going to be able to do this in most markets and still be able to decode the PSC.
The plus side is you have LOTS of transmitters to use/choose from so you can easily drive many of them to tune the model. So you should be able to get a very accurate model, but only for a somewhat limited range of signal levels because of the limitation noted above. Which still might be good enough for what you need the model for.

-Deuce-

[d0000h]

For this I use PC-Tel Scanner, this scanner measurement level of -120 dbm in fast scanning mode, this scanner don't realize SHO, this monitoring all PSC.
I try to use model tuning with PSC, for me is better because I can measurements a lot of signal for diferents trnasmitter.

Yes but if you are -105 from cell A, then you are probably measuring -70 from cell B or C or D, and can no longer decode your PSC from cell A which mean you cannot use the data point for model tuning (for cell A). That was the point of my first post.

[senjamir]

hi all,

is there anyone who has an experience on model tuning (GSM /UMTS 2100 ) , please PM(private message) me , its urgent requirement.


Regards,
Faib

I think this simple diagram might help you.

Br,

Sen

[senjamir]

I think this simple diagram might help you.

Br,

Sen

Here is the explanation of the procedure. Sorry for the delay. :)

1. Number of Models - You can decide with the number of models based on the the requirements. It can be one model for dense urban and one model for suburbs
2. Data Collection Method
a. TTM/TMR method - you will set up a transmitter and receiver
b. TEMS - collect data for model tuning using existing network with similar frequency (time saving and cost efficient )
3. Site Selection - this is depending on the scope of the project.
you can consider the following criteria
a. clutter - major clutter are being covered
b. 20m unobstructed antenna 15-30 antenna height
4. Drive Route Planning - 50 - 100 km of driving per site ; drive as many back roads and small streets
5. Data Post Processing - try to remove unwanted data and process the remaining good data.
6. Model Tuning in TCPU - (this is a long process and a tool that can be learned)
- main idea is to import the log files from DT and tune it based on the propagation model that is desirable from the parameters that you used.
Sample parameter:
a. Frequency
b. distance of cell to transmitter
c. height of transmitter antenna
d. height of receiver antenna
7. Results - just use auto tune and see to it that the standard deviation is as low as possible to have a better accuracy of data.

Br,

Sen

[ocsibacsi]

Yes but if you are -105 from cell A, then you are probably measuring -70 from cell B or C or D, and can no longer decode your PSC from cell A which mean you cannot use the data point for model tuning (for cell A). That was the point of my first post.
Yes, that could happen. (The same happened with the BSIC decoding in GSM. BSIC decoding didn't work when an adjacent channel was too strong, for example.)

Do you have real live experience with PSC measurement/decoding? Could a new Scrambling Code design help in this case?

[d0000h]

Yes, that could happen. (The same happened with the BSIC decoding in GSM. BSIC decoding didn't work when an adjacent channel was too strong, for example.)

Do you have real live experience with PSC measurement/decoding? Could a new Scrambling Code design help in this case?

PSC planning will not help. Because a good scanner should be able to decode the PSC even if Ec/No = -22 or so, you should be able to do a pretty decent model tuning drive on a live network, just not as good as with a dedicated CW transmitter/frequency.

I have never done model tuning on a live WCDMA network, only with a dedicated CW transmitter.

I have done it on live GSM, but as there is freq planning we could use a protected channel during our drives.

-Deuce-