Predicting and generating accurate coverage maps.

[quad_track]

It's a bit more difficult than that. There are limitations to the software that one may be using. For example, the software that I am required to use doesn't accept external building shape files. What software does?

Open source? qradiopredict uses FLightgear for 3D, although it's a very simplistic model with rather obsolete code.
For 5G in-building simulation for example, you need to have a 3D model with embedded metadata about type of materials used for the walls, as they all have different absorption and reflection coefficients. With ray tracing algorithms, you can see wave diffraction around structural elements and can visualise the results in 3D. Use Blender which is the best thing around for 3D stuff.
Implementing this in open source software is left as an exercise to the reader.

 

[paulears]

Forgive me, but you want to check a path, so you have to use what? Google Earth to show the area, then you build a facsimile in a 3D package, and then you can do it? er, how long would it take you to model the entire area the RF has to travel through, around, or be bounced off? Surely, while we can agree that it's possible, the time to build such a model to be accurate would be considerable?

 

[jonwienke]

Estimating coverage is always going to be inexact, given that seasonal foliage variations can have a significant impact on ground-level RF propagation, terrain models are simplified representations of the actual topography and generally don't consider factors like soil type and moisture levels (which also varies seasonally), and none of the models include accurate architectural drawings of all structures in the map area.

 

[quad_track]

Forgive me, but you want to check a path, so you have to use what? Google Earth to show the area, then you build a facsimile in a 3D package, and then you can do it? er, how long would it take you to model the entire area the RF has to travel through, around, or be bounced off? Surely, while we can agree that it's possible, the time to build such a model to be accurate would be considerable?

Google Earth is just a rendering engine. You don't have to use it as there are plenty other more advanced and open source.

Also, a wealth of data is available from composite sources: photogrametry, lidar, etc. Nobody in their right mind builds a 3D model of the whole earth from scratch. Look: https://www.youtube.com/watch?v=QcSSkWSE6ys https://www.youtube.com/watch?v=GrU846dieIM - this is Lidar in OsgEarth and can be automatically generated for large areas. https://www.youtube.com/watch?v=Z_7afNznnRs is OSM imported as 3D still in OsgEarth and it's available for the whole world. The only thing left to do is plug the radio propagation model of choice in that engine

 

[quad_track]

Estimating coverage is always going to be inexact, given that seasonal foliage variations can have a significant impact on ground-level RF propagation, terrain models are simplified representations of the actual topography and generally don't consider factors like soil type and moisture levels (which also varies seasonally), and none of the models include accurate architectural drawings of all structures in the map area.

I beg to differ. Spectrometry and near infrared imaging can give you enough information regarding foliage water content, and then you can derive seasonal variation, for example Corine has a high density map from which you can extrapolate all this information: https://sentinel.esa.int/web/sentinel/user-guides/sentinel-3-olci/applications/land-monitoring
Moreover you can use atmospheric pressure and humidity data not only for seasonal but also for daily variance. There's a Qgis plugin which will do this automatically for you, it's been written by an university in South Africa. (source: worked on a radio planning engine for a couple of years)

 

[zz0468]

This all sounds fascinating, but after a brief look at the qradiopredict link, it appears to be experimental at this time. And a great deal of what you're suggesting sounds like it's still vaporware.

Regardless of it's actual status of a real solution, it appears to be something of interest for those wishing to develop more accurate propagation prediction tools, and in that regard, it's quite interesting and shows a lot of promise. It's gratifying to see progress being made in coverage prediction software. It had leveled out for quite some time.

On the other hand, for those who are interested in building radio systems and use such tools as a means to an end, they're not nearly ready for prime time until the developers integrate all those promising features into a ready to run package.

Open source? qradiopredict uses FLightgear for 3D, although it's a very simplistic model with rather obsolete code.
For 5G in-building simulation for example, you need to have a 3D model with embedded metadata about type of materials used for the walls, as they all have different absorption and reflection coefficients. With ray tracing algorithms, you can see wave diffraction around structural elements and can visualise the results in 3D. Use Blender which is the best thing around for 3D stuff.
Implementing this in open source software is left as an exercise to the reader.

 

[quad_track]

This all sounds fascinating, but after a brief look at the qradiopredict link, it appears to be experimental at this time. And a great deal of what you're suggesting sounds like it's still vaporware.

Oh yeah, right, vaporware I suppose nobody implemented these algorithms: IEEE Xplore Full-Text PDF: in software, right? Could this be the normal useless stuff you find on IEEE?
That's just one of a few hundred papers about it by the way, because you seem to think that I'm just suggesting this and there's no functional code with it.
But of course people would like this kind of software for free if possible

In my opinion, for most amateur radio applications, Radio Mobile is very good software and is sufficient. Anything more advanced? Plenty of datasets out there, get familiar with GIS, GRASS, GDAL and dig the interwebs for the huge ecosystem of free software libraries that will do it for you.

 

[SCPD]

I will admit right off that I am not at all familiar with QRadioPredict; that said I went looking.
.
It appears that the designers have discontinued its development-- have I read this correctly?
.
Never-the-less, I am intrigued by the program and I will see if I can run it side-by-side against one of ours next week.... I'm willing to be pleasantly surprised.
.
Until then, I'm going to side with ZZ, echo'ing skepticism as to its real world performance.
.
.
.......................CF
.
.
(IEEE member, btw- fwiw.... )

 

[quad_track]

You might be looking in the wrong place, because I saw a pull request recently. However qradiopredict by itself is nothing special, just plain old Longley-Rice and some (very primitive) ray tracing. Its only redeeming feature was the fact that it could offload the calculations to an external 3D engine. AFAIK Flightgear was the first open source 3D engine to have radio signal calculations. You could walk around in 3D with a mock instrument and get signal estimation almost real time (as real time as the Linux kernel would let you ). You can read about it here: Radio propagation - FlightGear wiki

Nowadays that kind of stuff is old, as it came out more than 8 years ago. State of the art in 3D mapping is OsgEarth (from open source offerings) and offloading the ray tracing to GPU via Cuda. That kind of code is sometimes used in 5G aspects, especially for building surveys (there's a lot of research done here and there are literally thousands of people working on the whole software stack). Can't provide more details because NDAs.

Edit: oh btw., I suppose everybody here knows about Q-Rap: Radio Planning Tool , that's why I didn't bother to mention it. Neat little piece of software, and I know about more than one fork with nice additional features.
Sometimes you have to go digging not only through the master branch of the repo but also other branches

 

[zz0468]

Oh yeah, right, vaporware I suppose nobody implemented these algorithms...

I'm not disputing anyone's white papers on the subject, or the research being done. But qradiopredict appears to be a project intended to enhance realism of navaids and comms for an open source flight simulator. Perhaps it has use as a propagation prediction application, but as it appears to be experimental at this time, it's not going to find acceptance in a professional setting where plots produced are used to influence engineering decisions of real projects.

In a previous post you mention meta data containing materials characteristics for calculating losses in buildings. I'm not aware of anyone actually doing that outside of a research setting, so in my mind it's vaporware. You also mentioned that implementation in open source software is left up to the reader...

That's vaporware.

I stand by my previous skeptical comment that we mere mortals can only calculate so many RF paths for plotting coverage. But mother nature knows them all. RF coverage prediction will always have an element of uncertainty, no matter how good the algorithms are.

I'm a busy man... when it comes out as a commercial product where I don't have to spend time building a detailed 3d environment, I'll definitely be quite interested.

 

[wb6uqa]

Line of Site

Double the antenna height, then take the square root. Mt Wilson is at 5,000 feet, line of sight is 100 miles.

 

[paulears]

Ah, but the line of sight from Mt Wilson to 'where?"

 

[DaveNF2G]

Double the antenna height, then take the square root. Mt Wilson is at 5,000 feet, line of sight is 100 miles.

Line of sight or radio horizon? Slightly different things.

 

[jonwienke]

In order to be more than a rough estimate, a coverage map calculator needs ALL of the following:

1. Detailed topography map, including not just the surface coutours, but also soil type and current moisture levels and conductance. The latter can be guesstimated from recent weather, but there is little actual data available to put in a model.

2. Current vegetation conditions, including the species, height and foliation level of wooded areas (and the resulting RF reflection, diffusion, and absorption profile), with a high degree of granularity. This can be guesstimated in part from historical satellite data, but realtime data is not generally available.

3. Detailed and accurate 3D architectural models of all structures in the coverage map area which include all materials used in the building, and their RF characteristics. This data is generally not available.

4. Current weather conditions--temperature, pressure, humidity, precipitation, cloud cover, ionosphere, etc. This data exists, but isn't necessarily currently available in realtime outside the meteorology community.

Item 1 is partially available in the form of actual data. Item 2 can only be roughly guesstimated, unless you can construct accurate 3D renderings of each tree. Item 3 is not available, and is unlikely to be available any time soon. Item 4 is mostly available, if you can arrange access to realtime radar and other weather data.

Available data can result in an approximation that is accurate enough to be useful, but I don't see it being "precise" any time soon.

 

[jim202]

When I was working for the different cellular companies, we used a program called "TAP" that I believe was sold by Soft Write". It used Geodetic terrain files to provide the topography information. Then they had different modules to fit your application.

It was not cheap, but did a great job in providing coverage. Doing drive testing to verify the results the only thing we had to do was find the magic number for the foliage attenuation for our area. That only took the first drive around to verify we had it correct.

 

[zz0468]

When I was working for the different cellular companies, we used a program called "TAP" that I believe was sold by Soft Write". It used Geodetic terrain files to provide the topography information. Then they had different modules to fit your application.

TAP is a very good program, and it does a pretty decent job. But typical of all those programs, those fudge-factors for the various losses can change day to day, and it just isn't possible to map them all.

You end up getting a good understanding of what coverage looks like, but there will be areas where it is spot-on, and others where you missed it by 15 dB.

Some programs let you go in and make localized adjustments so it can all be documented on the map, but at that point the map is no longer purely predictive, it's just a handy way of documenting your field measurements.

 

[SCPD]

The science behind this discussion is fascinating.. and this morning I asked one of our grad students if she had time, would she run the "QRadio" program and let me know how it compares to some of our's. She's quite busy so I'm not holding my breathe- but enuff has been said about this in here, and keeping an open mind- I am scientifically curious,------- if nothing else.
.
One thing I'm not is a Model'er... I find its far easier to gather the data and write an explanation of what was observed rather than try and explain why the math didn't fit the observation.
.
.
That said- a brief story--
.
.
I had a project where we were testing a particularly high frequency microwave circuit over a range of approximately 50 miles- pretty much rolling desert- mesas, ravines... arid Southwest desert. The beam path was to skim at the ground surface, as low as possible.
What seem'd like on a regular weekly basis, signal strengths would drop precipitously, stay down a day or so, and return to normal. We looked at all the 'official data"- weather conditions, etc. and could not account for it at all. Need less to say, any models we had could not account for this weekly phenomenon.
Curious as exactly what might lay in that path, I drove out across the desert, following a lot of old ranch roads into the slick rock country. I chose a time to do this during the signal minimum, hoping that whatever was causing the effect would be out there... what ever it might be.... (I was psych'd to come across a temporary space alien base..... )
.
In a low lying playa , I came across some cowboys driving a large herd of cattle to the lake in this flat. I asked them how often they did this....
.
"we bring them here every week, Ma'm"
.
Ok, -- Cattle Attenuation.... Cool!... I guess............
.
I couldn't come up with anything else.. maybe cattle flatulence ??
Anyway, there was no code for Bovine Attenuation. When I got back to 'civilisation" I comment'd on the lack of this aspect in any program. Maybe this would be the basis of an interesting conference paper? I suggested it to my friend...
.
"Yes, that would be great, Lauri....," she said.. "YOU present it !... "
.
.
.......................CF

 

[zz0468]

...Anyway, there was no code for Bovine Attenuation...

See? Mother nature knows all about cows, including their RF loss characteristics. Probably their dielectric constat, too.

I know of a similar story with an AT&T microwave link up in the Owens Valley, where there was a regular cycle of long deep fades. That turned out to be trucks parking in a particular wide spot along the hwy 395, where path clearance was just above the road.

You can't map that stuff!

 

[jim202]

It is poor practice to have a microwave path skim close to the ground. You need to make sure you have more than enough clearance to allow for things moving around on the ground and account for trees.

Here in the New Orleans area, we get fog at certain times of the year that raises all sorts of problems with microwave paths. The best thing to do is get above the fog and any other obstructions. If it isn't fog, during the day, we get thermal inversion issues that causes the microwave beam to bend, or you get reflection issues with bodies of water.

On one 6 GHz. path that was 27 miles long, we used space diversity on the far end that ended up with the prime and secondary dishes almost 200 feet apart. The distance was found by moving the space diversity dish a number of times until the sweat spot was found and the path stayed working. The far end was out in swamp country.

Trying to engineer a microwave path is not all computer design. Some of it ends up being by the seat of your pants based on experience in the area the path will follow.

 

[Project25_MASTR]

The science behind this discussion is fascinating.. and this morning I asked one of our grad students if she had time, would she run the "QRadio" program and let me know how it compares to some of our's. She's quite busy so I'm not holding my breathe- but enuff has been said about this in here, and keeping an open mind- I am scientifically curious,------- if nothing else.
.
One thing I'm not is a Model'er... I find its far easier to gather the data and write an explanation of what was observed rather than try and explain why the math didn't fit the observation.
.
.
That said- a brief story--
.
.
I had a project where we were testing a particularly high frequency microwave circuit over a range of approximately 50 miles- pretty much rolling desert- mesas, ravines... arid Southwest desert. The beam path was to skim at the ground surface, as low as possible.
What seem'd like on a regular weekly basis, signal strengths would drop precipitously, stay down a day or so, and return to normal. We looked at all the 'official data"- weather conditions, etc. and could not account for it at all. Need less to say, any models we had could not account for this weekly phenomenon.
Curious as exactly what might lay in that path, I drove out across the desert, following a lot of old ranch roads into the slick rock country. I chose a time to do this during the signal minimum, hoping that whatever was causing the effect would be out there... what ever it might be.... (I was psych'd to come across a temporary space alien base..... )
.
In a low lying playa , I came across some cowboys driving a large herd of cattle to the lake in this flat. I asked them how often they did this....
.
"we bring them here every week, Ma'm"
.
Ok, -- Cattle Attenuation.... Cool!... I guess............
.
I couldn't come up with anything else.. maybe cattle flatulence ??
Anyway, there was no code for Bovine Attenuation. When I got back to 'civilisation" I comment'd on the lack of this aspect in any program. Maybe this would be the basis of an interesting conference paper? I suggested it to my friend...
.
"Yes, that would be great, Lauri....," she said.. "YOU present it !... "
.
.
.......................CF

Back when I was working out in the West Texas oilfields...my boss would've strung me up from a tower for not checking for a minimum clearance of the first Fresnel zone.

I've got a couple of those stories actually. One involved a proposed link for a new tank battery (or was it a gas plant). Told where we could put the tower and were told how high we could go...cleared the first FZ and everything looked really good on paper. Then I said, let's drive it (I would've preferred flying but at one point the path was only 35 feet off the ground). Wouldn't you know it...a pine tree (only tree for 5 miles in any direction) was right at that low spot. The other involved an established link in a field that someone approved a tank battery in the path 100 feet from the tower. Of course, we didn't find out someone had built a tank battery in the path until we drove (3 hours one way) out to find out why the link had gone down.

It is poor practice to have a microwave path skim close to the ground. You need to make sure you have more than enough clearance to allow for things moving around on the ground and account for trees.

Here in the New Orleans area, we get fog at certain times of the year that raises all sorts of problems with microwave paths. The best thing to do is get above the fog and any other obstructions. If it isn't fog, during the day, we get thermal inversion issues that causes the microwave beam to bend, or you get reflection issues with bodies of water.

On one 6 GHz. path that was 27 miles long, we used space diversity on the far end that ended up with the prime and secondary dishes almost 200 feet apart. The distance was found by moving the space diversity dish a number of times until the sweat spot was found and the path stayed working. The far end was out in swamp country.

Trying to engineer a microwave path is not all computer design. Some of it ends up being by the seat of your pants based on experience in the area the path will follow.

Fresnel Zone clearance was one thing one of my mentors loved to preach, especially for backhauls. He cared enough to teach me about it that I still check it. I actually had never dealt with diversity receive until just recently. A link got upgraded to handle the additional bandwidth of a MCC7500 along with site traffic and they added diversity receive antennas. I thought it was a bit weird as the shot isn't over water but I'll roll with it. One of my mentors used to tell me about his early days working in GOM about using diversity receive to counter the tide on off-shore rigs but I haven't seen it much since.