Path Loss and Coverage Model Simulation

[ChristianH]

Hi,

i am currently working in a path loss propagation package. The package calculates the signal loss for 20Mhz-20Ghz transmissions, based on the Longley Rice Model and an upgraded Irregular Terrain with Obstructions Model. The calculation is not "grown on my dung". Its origin is a model published in a thesis in the 70ies and written in Fortran. To keep long things short, I ported the c code (made in the early 2000's) to lazarus and added a class which reads the Shuttle Radar Topography Mission based topographic information in order to do a proper calculation with real elevation information. Further on I also wrote a simple renderer (based on BGRABitmap Library) which draws the data and the topographic map in an image. Since the Digital Elevation Data looks like a "fractal" it requires some "special" handling to look nice. The first thing I made was adding some hill shading (https://pro.arcgis.com/de/pro-app/latest/tool-reference/3d-analyst/how-hillshade-works.htm) and then I put some gradient coloring code to it.

The result of this currently looks like this:
https://github.com/TetrisSQC/pasitwom/blob/main/output.png

You will find the complete project here:
https://github.com/TetrisSQC/pasitwom

The next step is to paint the map using LazMapViewer. I did not found time to dig deeper into this project, so I kindly ask if somebody has the knowledge if and how I can draw an image with known minimum and maximum latitude and longitude coordinates over an existing map.

Christian

[MarkMLl]

Congratulations, that's a big job. I've never tried it myself but I've been on the receiving end of those sort of propagation calculations: MPT1411 paths with effing big lumps of granite in the paths.

MarkMLl

[ChristianH]

Thanks, yes I was aware that this will take a while to assemble something - especially since the last time I heard of path loss was in signal theory 21 years ago. Nevertheless it was great to read the papers, mostly because of the cryptographic function names zlsq1 (redo LOS horizons), ascat (scatter attenuation)...
There are a few things left, like adding a license (which I think will be LGPL). I hate to adorn myself with other people's feathers, especially since the only achievement in the case is mostly the compilation of the work of third parties. But I learned a lot and it was quite "fun", in a nerdy way.
Oh you could also define the Antenna pattern if necessary. This part of the code has been ripped of the splat!hd (https://www.qsl.net/kd2bd/splat.html) project. I might have to check the results of splat, compared with my code. I'm especially interested in, if the performance will be different. I split the calculation into 4 areas and do them in 4 threads. A complex model simulation on my iMac takes about 15 minutes and on my Arm based Air 5 minutes (which is kind of impressive). Anyhow I assume it can be improved in points of performance by using a faster math library.

Christian