[postgis-users] RE: MI-L Off Topic: ham radio modeling over open source

Juanse georef at tie.cl
Wed Feb 19 09:48:21 PST 2003


1.- That was a good answer, that I really appreciatte, and as I am not an
engenieer just a geographer (that skiped because of work abroad the last
physics module at university: Ondas, Waves?) so no idea at the moment what
is the implication of the up to 14 dB of error in our arrangement. I will
have to discuss with the rest of the team which way to proceed.

2.- I think my question and the answer might bother other members of the
lists, so we might keep the comunications between us and I summaryze at the

3.- My basic question is still open, how can i get USGS DEM or SDTS DEM data
starting with topographic (contour lines) data, having at hand
postgis/mapserver, mapinfo/verticall?. Basically all the softwares seen uses
as input data for modeling terrain DEM in USGS data format.


----- Original Message -----
From: Cameron Crum <c-crum at waveconceptsintl.com>
To: <Jerry.Bratton at alltel.com>
Cc: <georef at tie.cl>; <mapinfo-l at lists.directionsmag.com>;
<mapserver-es at yahoogroups.com>; <mapserver-users at lists.gis.umn.edu>;
<postgis-users at postgis.refractions.net>
Sent: Wednesday, February 19, 2003 1:22 PM
Subject: Re: MI-L Off Topic: ham radio modeling over open source

> First the software which used to be located at CPLUS.org is Radio Mobile
Freeware by VE2DBE and is still at the site
http://www.cplus.org/rmw/english1.html. It is different from the
RadioSoft-RadioLink program mentioned in your follow up e-mail. The
RadioLink program  is a point to point software only. Second, the Radio
Mobile Freeware program DOES NOT perform predictions just
> like dBPlanner or any other program with the exception of EDX now owned by
Comarco. Radio Mobile and EDX both utilize the ITM model. A bit of history
here... the ITM (Irregular Terrain Model) is basically Longley-Rice with
some modifactions made by George Hufford at the ITS in Boulder, CO. For
those not up on propagation models, this was the de-facto standard in
> modeling reccomended by the FCC until a few years ago. It predicts with a
typical standard deviation of 10-14 dB. Longley Rice is a semi-empirical
model and does not require feeding in drive test data to "calibrate" it.
Most other tools on the market use a completely impiracal slope-intercept
model like Cost231, Hata, Lee, Okumura, or derviations of these. These
models are
> simply far-field, straight line models that are "tuned" by going out and
setting up test transmitters, driving around and collecting signal
strengths, then inputting these into a tuning algorithm which makes
adjustments to the slope and intercept points until the line "best-fits" the
collection of data. Lump sum losses can also be added in "post-prediction"
to try to make up
> for things like clutter losses. These models, depending on the engineer
who tunes them, can have accuracies as low as 7 dB, but this is rare.
Typically, standard deviations are from 9-14 dB. To achieve lower, it
requires extreme care in setting up and conducting the drive test, and to be
accurate for every site in a system, a seperate drive test must be conducted
for each
> site. In my opinion, if you are going to drive every site, why do you then
want to go and approximate what you already know is there. It doesn't make
much sense, but then most people who use these models obviously are not too
concerned with accuracy or simply don't know any better.
> Now, I'm no dBPlanner fan, and in fact am a direct competitor to them, but
I feel that I must defend them on this matter. dBPlanner, in its infancy,
used only one model, the CRC-PREDICT created by Dr. James Whittaker of the
Communications Research Center in Canada. This model is a truly predictive
model based on the theories of physical optics (Freznell-Kirchoff wave front
> base on Huygens Pricliple for those who care). It predicts with typical
accuracies of 5dB standard deviation or better when used with high resoution
(30m) terrain and clutter databases. It requires no drive testing for
"calibration". There are only two tools in the world which currently use
this model, dBPlanner and our tool Athena. dBPlanner has since sold
themselves short,
> and inspite of hiring Dr. Whitaker away from the CRC to help them, they
went ahead and implemented several slope-intercept models as well. They
still call the CRC-PREDICT their "flag ship" model, but the others are
available if one chooses to use them.
> Now, if you are still reading, you may be wondering why others have not
implemented the CRC model if it is truly superior. There are several
reasons. First, most of the RF tools on the market were written by
programmers and not engineers. It is doubtful that there are even people
within these organizations who even understand the true nature of
propagation modeling or could
> explain what the differences are in modeling techniques. Second, it is
expensive to reprogram a tool to use such a model. It requires knowing what
the model is doing which brings us back to the first point. Third, everyone
sells their tool as the most accurate (especailly if the client buys some
additional engineering services to drive test and tune the models). If these
> companies start selling a tool which doesn't require drive testing, they
are going to loose a lot of revenue. Fourth, if they admit the CRC is a
better solution, their clients may jump to one of the tools who already
implements it instead of waiting around for them to implement it. Fifth, and
perhaps most important, Marconi has arranged an exclusive licencing
arrangement with
> the CRC so nobody else can license source code. Since our license predates
their license and we have a sub-licensalbe source code license, we are
currently the only source in the world to obtain the model. We don't give it
away and most of these companies won't pay for a model despite the accuracy.
> So, to say the freeware provided in the previously mentioned links
predicts just like the "expensive" tools is incorrect. However, if I didn't
have a tool and were looking to run some predictions on one or two sites,
the Radio Mobile Freeware would not be a bad choice. I would trust its
results over an uncalibrated slope intercept model any day. My suggestion
would be to use
> as high a resulotion terrain as it allows...currently the SRTM data which
itself is quite expensive outside the US. Also, the program does not account
for clutter which can have dramatic effects on the propagation attenuation.
So, my advise is....be careful in depending on these results for any high
level engineering.
> Cameron Crum
> Jerry.Bratton at alltel.com wrote:
> >         There used to be a freeware utility @ CPLUS.Org I went there
today and it looks like a different company. This program lets you bring in
DEM files and performs radio predictions just like the expensive programs
like dB Planner. Luckily I saved the program and the website for future
reference. I can upload it to my FTP server if you would like to try it out.
> >
> >         Jerry
> >
> > -----Original Message-----
> > From: Juanse [mailto:georef at tie.cl]
> > Sent: Tuesday, February 18, 2003 12:19 AM
> > To: MapInfo; mapserver-es at yahoogroups.com;
> > mapserver-users at lists.gis.umn.edu; postgis-users at postgis.refractions.net
> > Subject: MI-L Off Topic: ham radio modeling over open source
> >
> > Hi every one and sorry before hand for this off topic requirement
> >
> > Under a non profit organization sponsoring we are developing an internet
access system for isolated school based on ham radio data transmition. On
this effort we are looking for any info regarding the siting of antenas
/repeater. We would like to explore the modeling of the signal taking care
of topography to define the best arquitecture to supply all school at least
> >
> > We have 1:50.000 digital topographic maps of the study area, and it
available either through mapinfo/vertical mapper or postgis/mapserver.
> >
> > Under our preliminary research there are a couple of comertial software
on the range of US$500 to US$5000, and some basic possibilities under
opensource (like splat http://www.qsl.net/kd2bd/splat.html).
> > SPLAT! imports topographic data in the form of SPLAT Data Files (SDFs)
that may be generated from a
> >
> > number of information sources. In the United States, SPLAT Data Files
are most often derived from U.S.
> >
> > Geological Survey Digital Elevation Models (DEMs) using the usgs2sdf
utility included with SPLAT!.
> >
> > Any one know of  tools to derive DEM like the ones need by splat SDF
from contour lines?
> >
> > Any one knows of other open source tool that could be usefull for our
> >
> > Sincerely
> >
> > Juanse
> >
> > temuko-Chile
> >
> > ---
> > Outgoing mail is certified Virus Free.
> > Checked by AVG anti-virus system (http://www.grisoft.com).
> > Version: 6.0.449 / Virus Database: 251 - Release Date: 27/01/03
> >
> > ---------------------------------------------------------------------
> > List hosting provided by Directions Magazine | www.directionsmag.com |
> > To unsubscribe, e-mail: mapinfo-l-unsubscribe at lists.directionsmag.com
> > For additional commands, e-mail: mapinfo-l-help at lists.directionsmag.com
> > Message number: 5572
> ---------------------------------------------------------------------
> List hosting provided by Directions Magazine | www.directionsmag.com |
> To unsubscribe, e-mail: mapinfo-l-unsubscribe at lists.directionsmag.com
> For additional commands, e-mail: mapinfo-l-help at lists.directionsmag.com
> Message number: 5579

Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.449 / Virus Database: 251 - Release Date: 27/01/03

More information about the postgis-users mailing list