View Full Version : Model validation (urgent)
justdream
2008-11-13, 05:16 AM
To all Model tuning guys,
How could you accurately validate your model?
I need to hear from you in details ASAP (tonight will be very good :)as I'm very far behind schedule)
raedsab
2008-11-13, 03:46 PM
http://rapidshare.com/files/163285565/Model_Calibration.pdf.html
justdream
2008-11-14, 04:33 AM
Thanks to all my friends,
I think you didn't understand my inquiry: suppose that you have created a model with good STD and it gives you good coverage prediction
But how could you ensure that this coverage similar to the real situation?
According to my information we should examine the difference between the predicted signal and the measured one………ok
Now what's the type of the measured signal which will be used to validate the model:
CW "unmodulated" or modulated signal collected from real sites "in case of modulated signal, will we collect it by scanner 'scanning mode' or by mobile handset 'in case of mobile handset will we use dedicated or idle mode'?
- This subject is very important for me, so I need to know what did you do in your previous model tuning validations and if you have document about this subject it will be highly appreciated,
Please help me to answer these inquiries, I know that I bothered all of you about model tuning but right knowledge and experience take long time of searching and reading.
- I will be very happy if I could add you to my contacts at skype to discus in more details, so if you don't mind PM me.
Once again, thanks a lot.
hassen
2008-11-14, 05:30 AM
http://rapidshare.com/files/163285565/Model_Calibration.pdf.html
hi friend
do you have like the same doc that you just post it on forum from e******* TCPU?
hassen
2008-11-14, 05:45 AM
Thanks to all my friends,
I think you didn't understand my inquiry: suppose that you have created a model with good STD and it gives you good coverage prediction
But how could you ensure that this coverage similar to the real situation?
According to my information we should examine the difference between the predicted signal and the measured one………ok
Now what's the type of the measured signal which will be used to validate the model:
CW "unmodulated" or modulated signal collected from real sites "in case of modulated signal, will we collect it by scanner 'scanning mode' or by mobile handset 'in case of mobile handset will we use dedicated or idle mode'?
- This subject is very important for me, so I need to know what did you do in your previous model tuning validations and if you have document about this subject it will be highly appreciated,
Please help me to answer these inquiries, I know that I bothered all of you about model tuning but right knowledge and experience take long time of searching and reading.
- I will be very happy if I could add you to my contacts at skype to discus in more details, so if you don't mind PM me.
Once again, thanks a lot.
Hi
make Tems investigation for the first time in Idle mode to ovoid power control and dtx feature ,by choosing one site for example ,and TEMS locked at a frequency of a cell belong your chosen site .do it for each clutter the same method of measurement,and finally made a comparison between measured value and predicted values gived by tool.with atoll is possible to make comparison.
Gute Nacht Freund
Hassen
justdream
2008-11-14, 06:28 PM
Hi
make Tems investigation for the first time in Idle mode to ovoid power control and dtx feature ,by choosing one site for example ,and TEMS locked at a frequency of a cell belong your chosen site .do it for each clutter the same method of measurement,and finally made a comparison between measured value and predicted values gived by tool.with atoll is possible to make comparison.
Gute Nacht Freund
Hassen
<!-- / message -->
Thanks my Friend,
I already made that for 'X' area by TEMS using Mobile handset in idle mode locked to sites SC but my customer get another drive test made by scanner for the same 'x' area which gave him more optimistic results than mine,
//(I think in case od scanner they didn't lock to any SC)//
so which one is the accurate one or which one I should use to calculate the accuracy of my model?
Thanks a lot for your concern.
depending on the calibration and postprocessing I would go for the fast scanning receiver (or scanner as you call it). :D
justdream
2008-11-14, 11:20 PM
Dear DCA,
could you explain in more details what do you mean by:
'depending on the calibration and postprocessing I would go for the fast scanning receiver'?
for serious fieldstrenght measurements it takes more then just record RSSI values. the installation in your car needs to be calibrated for example . For the postprocessing at least use Lee averaged values.
But there more things to consider LOS/NLOS and so one. It would take to write a manual to get you implement the correct values ;)
justdream
2008-11-15, 06:48 PM
It would take to write a manual to get you implement the correct values
yes, Boss, could you help?
thanks in advance
roots
2011-02-01, 06:24 PM
http://rapidshare.com/files/163285565/Model_Calibration.pdf.html
Can anyone reupload this file Please :(
vBulletin® v4.2.5, Copyright ©2000-2024, Jelsoft Enterprises Ltd.