Just for Experts,
I build a model tuning for UMTS and I get good STD (7.8 or 8) but when I run coverage prediction to test the model, I got disontinuty in the coverage besdie limited high signal strength around the site as shown in the figure, beside the lovely circle around the site where’s the arrow pointing
Who could advise?
when I let the ASSET tool calculate the clutter through loss automatically,
the STD reduced as shown
but the covergae prediction became very bad
how come reducing STD badly affect the coverage prediction?
Low STD mean good model,
what do you think guys?
That is due to the deep valley of antenna pattern.
using through loss and let it run “wild” automatically is a dangerous practice. It will get a model fit the available data points mathematically best, but not necessarily making RF-sense.
firstly, thanks my friend for your concern.
ok,
now to edit clutter through manually, what’s the main idea?
could you advice ASAP with example?
Thanks
I don’t have real example. But if a clutter type (like forest, tall building) is likely to block the way between MS & BS, then you would give it a higher loss. If it pass through water, then you would actually “gain” rather than lose. …
You could also wonder if your DT data is covering all the clutter classes you are using/tuning with enough valid samples but you probably looked at that already 
Hi friends.
From my previous experience on model tuning. I am just leaving through loss “zero”. Let the Clutter Offsets play around, it has a bigger impact on the accuracy of your prediction.
Your STD is entirely great. Have you excluded bins lying on clutter which less than the required amount? How many clutter did you include on tuning?
The more ugly your prediction is, the better. 
From your prediction, I think it is quite normal. Optima2 is correct, your terrain might have an influence. Terrain might be going steep with respect to the site.
BR,
Anne
Hello all,
There are some parameters to consider to fine tune your Propagation Model, first I want to ask if how is clutter data distributed?, Is the clutter map resolution the same as the height data?. Refer to the attached image to see some of the values i use in my Model:
The Clutter Offsets values are the critical values and will have big impact on your predication coverage and as anita_2k mentioned the ‘Through Loss values’ are all “Zero”. ‘large Loss values’ will increase the attenuation
in that clutter section. I have tested a value of 20 in ‘Through Loss’ and in Clutter Offset and the coverage was zero in that clutter section.
On thing also to consider (Use Clutter Height) check box, this box if checked will take into consideration the Building height data (If any) and will increase the array creation time. Buildings will be considered here as obsticales.
Dear DCA,
you are right, my DT doesn’t cover all the Clutter classes
I’m tuning using 6 sites which contain bins of Open, Denseurban, Induscomercial, parks, Blockbuildings, denseblockbuild an residhigh (each class contains more than 1000 bins for dense urban and 700 bins for urban)
I couldn’t use mor ethan 6 sites, otherwise my STD will get out of accepted range
so if you are in my position, what should you do to get good STD and real coverage prediction
Fast response will be highly appreciated
you may need more data
The remainig data I have is not good and no way to collect new data,
what happened is happened, I want now to get the optimum result.
I will summarize my inquiries as I don’t have more than two days:
you told me guys to edit clutter offset and leave through loss as zeros, are you sure??? I didn’t read that in Aircom doc.
I feel that my EIRP (43 dBm) is not good, what do you see from your previous experience with UMTS model tuning?
To validate your model, could I validate it by TEMS collected data or it must be CW data?
Again fast response will be highly appreciated.
It looks like you only have data for specific clutterclasses and the question is if you have enough data to get a more accurate result.
Then you need more data to tune the rest also.
About the through loss, for your purpose, I think you don’t need it.
And looking at your result, it’s not that bad. Probably enough for a ’ quick ’ job
if you see my work as not bad, so please be my customer and accept the project
but my friend my prediction shows little high signal stregnth (blue one -20 to -65), this is not logic
I like your Guidence be continue in future.
Guys,
what about validation? How could I validate my model?
Hi Juststream, I got your PM regarding “deep valley”. I cannot attach the picture with PM. When I said “deep valley”, I was talking about vertical antenna pattern – see where the green arrow is pointing. And because of that, there are certain area near the base station will experience weaker signals than the area a bit further
And because of that, there are certain area near the base station will experience weaker signals than the area a bit further
Do you think my friend that these circles could be found in real?
also do you think that Near field would help in removing these circles?
Thanks in advance.
