Predicting and generating accurate coverage maps.

Status
Not open for further replies.

N3KGD

Member
Joined
Sep 30, 2007
Messages
305
Location
Erie, PA
Hello all,

I've been playing around with Radio Mobile for years to generate coverage maps. I don't think I'm doing everything just quite right though. I know radio coverage can never be 100% guaranteed per-se, however, I've been tinkering around with the software tonight. So, I went ahead and added my base, mobile and portables to the system I am working on and inputted all values including coax line loss, antenna gain (if any, or negative in this case for portables), antenna height, etc...

Here's the thing, I can make these maps lie like you wouldn't believe exaggerating coverage where coverage isn't at or I can make them so tight that it doesn't display coverage where coverage is at. I understand that coverage maps are estimates and cannot predict exact coverage due to atmospheric conditions amongst several other variables.

Here's what I need to know: I want to try and get a generally accurate coverage map for my base radio system.

When I am in the network window, I can select the following modes of variability: spot, accidental, mobile and broadcast and adjust percentages such as time, location and situations. Then, when it comes to generating the coverage map, I can adjust the thresholds for the map coverage depending upon the method used to calculate the map: S-Unit, dBm, uV, dBuV/m.

I've went ahead and made some maps and did field testing to compare. Some were a little exaggerated while others were too tight. What are some good settings to get a general day-to-day idea of coverage? I'm having a hard time deciphering the dBm also. I want to be able to hear radio traffic on portables. Doesn't have to be clear, but I don't want it being practically all noise to the point where I have to put headphones on to understand what the person is saying either in a fringe zone. What is a good "usable signal strength" value in dBm or otherwise? Furthermore, what are good values to select for modes of variability and the percentages therof.

Thanks,
Nick
 

Thunderknight

Member
Premium Subscriber
Joined
Jan 31, 2008
Messages
2,216
Location
Bletchley Park
The value is different based on the frequency, analog vs digital, and how clear you expect the signal to be. Values are detailed in TSB-88.
For vhf or uhf, you can try around a -106dBm for a reasonable analog signal with commercial grade radios. But it does vary a bit based on the radio being used.
 

zz0468

QRT
Banned
Joined
Feb 6, 2007
Messages
6,034
It's been a long time since I played with Radio Mobile, but I use a professional coverage plotting program on an almost daily basis. I can offer up a few comments that may or may not help make more accurate maps, but they will help you understand that making accurate maps is going to be close to impossible.

First off, Radio Mobile uses the Longley-Rice model, which is a mathematical model of propagation, and the set of formulas the program actually uses to make it's calculations (to put it in simple terms). The programs I use have multiple models to select from, depending on the conditions.

For example, I've found that the Bullington model is more accurate than Longley-Rice in open conditions where there aren't a lot of man-made structures.

Okimura is a model that was specifically developed for dense urban environments.

Another issue that you will run into is that of antenna patterns. Particularly the vertical pattern. Unless you have an antenna range, getting an accurate model of the antenna pattern is close to impossible, but that will have a profound impact upon the quality of the coverage map produced. Over the years, I've been finding it more useful to use an isotropic antenna so I can model what the site is capable of, since modeling the actual antenna in it's actual installed conditions is virtually impossible. Undocumented lobes and nulls in the actual antenna pattern will skew the plot vs. measured results, and you'll never understand why.

The idea of running a plot, then drive-testing it and making measurements is a good one, but you will find (as you did) places where the plots are either overly optimistic or overly pessimistic. The root of the problem there is that, while the software only knows so many ways of calculating out the coverage, mother nature knows 'em all.

Radio Mobile doesn't know about each and every building or reflective or absorptive surface the signal will run into.

If your plots are splitting the difference evenly between too much and too little, you probably have the program parameters optimized about as good as you can.

If you need to make minor tweaks here or there to plot specific neighborhoods, you can add or subtract whatever field margin loss parameters there are to fiddle with until the signal levels work out, but what works in one area, may not work right in another.

...I've went ahead and made some maps and did field testing to compare. Some were a little exaggerated while others were too tight. What are some good settings to get a general day-to-day idea of coverage?

I can't recommend specific values, but adjust things and see what gets you close to actual measurements. I'd start with a map of the desired area, and some field measurements you can take home with you with actual signal strength measurements.

Also, how are you getting accurate measurements? It does no good to compare if your measurements are flawed.

I'm having a hard time deciphering the dBm also.

How so? I find dBm to be useful when plotting out coverage because all you have to do is add and subtract the dB for transmitter power, antenna gains, losses, etc.

But microvolts is more useful when rating receiver performance.

I want to be able to hear radio traffic on portables. Doesn't have to be clear, but I don't want it being practically all noise to the point where I have to put headphones on to understand what the person is saying either in a fringe zone.

Is this an actual coverage issue, or a plotting issue? It is what it is, and plotting the coverage won't fix it. But it MAY help you understand why the coverage is poor in some areas.

What is a good "usable signal strength" value in dBm or otherwise?

Depends. If you're looking for mobile coverage, use the receiver sensitivity specification. Example - 0.25 uV for 12 dB SINAD. There are tables on the internet to convert from uV to dBm in a 50 ohm system. So taking that spec and converting it to dBm is a trivial task.

If you're wanting to plot portable coverage, you need to account for whether or not the portable is "on the hip", as well as the likely "negative gain" of the portable's antenna. Professionally, I add about 20 dB loss to account for portable on the hip coverage. Want indoor coverage? You have to account for that, too. What type of building? Wood structure? Commercial? High rise?

Plotting coverage can become a nightmare of variables, all of which need to be accounted for.

Furthermore, what are good values to select for modes of variability and the percentages therof.

As I said, it's been a while since I used Radio Mobile. One of the things I didn't like about it was the names of some of the terms were different from industry standards I was used to. But that said, the Longley-Rice model wants a percentage for "confidence", "time", and "location".

Use "confidence" of 50%, and 95% for time and location. I don't know if or what Radio Mobile calls for those parameters.

It can be helpful to read up about the Longley-Rice model, but it's pretty dry and beyond the scope of most people's interest or need.

I hope my rambling has been useful.
 
Last edited:

Project25_MASTR

Millennial Graying OBT Guy
Joined
Jun 16, 2013
Messages
4,164
Location
Texas
It really just depends. A lot of time it takes some trial and error to get going. I tend to plot using the ERP values of the SU equipment instead of ERP of the fixed point base. Helps provide more real world uplink values. I also have my plot set to assume a 15 dB difference in my RX signal levels (easy way to show mobile and portable coverage). So if I'm building a plot for 50W mobiles using 1/4 wave antennas I will set the stations ERP to 50W and typically measure out to -110 dBm with a 90%-90% probability.


Sent from my iPhone using Tapatalk
 

SCPD

QRT
Joined
Feb 24, 2001
Messages
0
Location
Virginia
Hi Nick.. :)
.
You are to be compliment'd on taking on such a study.... and I can't add much to what ZZ and the other have suggested.
.
I will add, for instance, The Longley–Rice Models were design for broadcast stations, where so many of the variables are known/controlled. Other programs are empirical math formulae's; adaptable to a scientific model but not easily translated to the practical day to day..... but you know that.
.
The last thing I want to inject here is a lot of bable- so I'll cut to the chase.
How necessary is it to know the exact numbers?.. If its purely for the sake of the science, I think your doing fine..... but translate that to the real-world; no wonder its getting jumbled.
.
Working in this arena I can spin you off any number of computer models-- the devil is, as so much in life, in the details.,,, and when we set up a remote site, I seldom ever use the science. We just go by what we think, and experience has taught us, will work.
.
I'm babbling, No?
.
Okay, here is "Coyote's Formulae" - maybe vastly over simplified, but the one I use for V/UHF all the time- especially when siting low power'd telemetry stations--
.
The distance in miles equals the square root of the height in feet of the sum of both antennas in the path, times 1.41
.
This is (almost) without regard to power levels.... giving me 99% contour of coverage.. and very simple to calculate.
.
.
Irregardless of what model you decide to use, keep experimenting ! .... :)
.
.
..................................CF
.
.
 
D

DaveNF2G

Guest
Okay, here is "Coyote's Formulae" - maybe vastly over simplified, but the one I use for V/UHF all the time- especially when siting low power'd telemetry stations--
.
The distance in miles equals the square root of the height in feet of the sum of both antennas in the path, times 1.41
.
This is (almost) without regard to power levels.... giving me 99% contour of coverage.. and very simple to calculate.
.

Interesting. I'm curious as to how you derived the formula. I recognize root2, of course, but maybe you could do a YouTube video about the rest. (I like watching Numberphile, btw.)
 

Project25_MASTR

Millennial Graying OBT Guy
Joined
Jun 16, 2013
Messages
4,164
Location
Texas
Interesting. I'm curious as to how you derived the formula. I recognize root2, of course, but maybe you could do a YouTube video about the rest. (I like watching Numberphile, btw.)



It's a modified LOS coverage assessment. I've actually got a GCT Level 2 manual that has something very similar. VHF it suggests multiplying by 1.3 instead of 1.41, UHF 1.2 and 800 1.1. Above 1 GHz you assume direct LOS is required.

Judging what I've read regarding CF's AO, I'd say that is a factor based on the flat terrain common to the high plains terrain found in parts of New Mexico, Colorado, Oklahoma and Texas.


Sent from my iPhone using Tapatalk
 

SCPD

QRT
Joined
Feb 24, 2001
Messages
0
Location
Virginia
Hi Guys--
.
That 'formulae,' if I can be so bold to call it one, is a simple calculation of a radio VHF/UHF horizon... a horizon consider'd slightly greater than visual line of sight.
.
And of course, there are many variables- like hills, mountains, --- there are environmentals like humidity and other frequency dependent detractors.
I am sure that you that have been in this field for any length of time, just "know' what will work... this simple equation puts it into simple numbers...... And it is just a guide.

.
If a signal is unusable because it is beyond that horizon, it will be unusable regardless -- (unless we now switch to other modes like scatter- but this is beyond "Coyote's Equation.") I have found using this 'formulae' that 90-99% of the time, if we are within the radio horizon, there will be a usable signal (- hills and mountains- bets go down ... ;) ). If it is initially too weak to be of use, I will "brass it" up to a higher power with beaming, going QRO etc... even going to passive repeating by bouncing a signal around a hill by beaming at a distant mountain side (one of my favorites.)
.
If my lab had a motto (or two) it would be that tired cliche "KISS".... so familiar to any techy's mind set.... for whether I am off a few Dbm's or not, it only appears in the written reports- not in the actual field performance- which is the ultimate measure of success.... for Simple trumps all.... :)
.
.
...................CF
 
Last edited:

zz0468

QRT
Banned
Joined
Feb 6, 2007
Messages
6,034
Hi Guys--
.
That 'formulae,' if I can be so bold to call it one, is a simple calculation of a radio VHF/UHF horizon... a horizon consider'd slightly greater than visual line of sight.

Adding to Coyote's comments...

The 1.41 in the formula is an expression of the approximate 4/3 earth radius that's used to account for the atmosphere in rf propagation where the variable index of refraction bends the signal toward the earth as it propagates outward. It'll eventually run into 'earth bulge' and hit the ground, but if the earth were 1/3 larger than it is, RF would just follow the curvature.

That number is refered to as 'k' factor, and in most propagation software, it's 4/3 or 1.33 under standard conditions. I'm not sure why the common formula shortcut uses 1.41. That equates to 4.2/3 earth radius, or something like that. Empirical data, probably. I've seen other values used, but the 1.41 seems to be the most common.

Some folks have had success using software like Radio Mobile and some of the commercial products simulating over the horizon weak signal propagation by using k factors as high as 25 (!).

And of course, there are many variables- like hills, mountains, --- there are environmentals like humidity and other frequency dependent detractors.

See how far you can talk, then work the formula backward and see what 'k' is that day. =)

I am sure that you that have been in this field for any length of time, just "know' what will work... this simple equation puts it into simple numbers...... And it is just a guide.

It's what all those engineers have been scribbling on napkins all that time. It works!
 

paulears

Member
Joined
Oct 14, 2015
Messages
769
Location
Lowestoft - UK
I suspect the problem is simply because the linearity and accuracy of the maths used to enable the prediction cannot take into account the almost random local issues once the antennas are not in free space. If you plot a path over a shortish range with two antennas at 50ft above ground and then change it to 100 ft above the ground, the programme works properly. It will take into account the presence or absence of a large hill but it cannot identify the 20 storey building that could be on that hill. Once you drop into what the radar people call ground clutter, it's a guess. Reflection, refraction, diffusion, absorption all impact greatly to the path loss calculations, but the software cannot predict these. If I use similar software and enter my current location, at a height above ground of 1m (sitting in front of the TV in my house), the software sees I am on top of a hill (I am) but ignores the huge absorption from the in leaf tree that is very wet today, 25ft to my right (the east), and ignores the two storey house blocking the path to the west. So entering say 4W, antenna gain of -2dB and height of 1m at this location suggests a fairly decent path - that is totally absent.

This is the problem with prediction at or near ground level. The programme could be totally accurate for a rolling countryside with nothing but grass - but it cannot cope with the real environment at low level, especially urban areas.
 

SCPD

QRT
Joined
Feb 24, 2001
Messages
0
Location
Virginia
I was going to add some more technical stuff to the discussion, but I think you-all have cover'd it marvelously.
_________________________________
.
I love anecdotes- for there is only so much I can glean from reading formulaes and doing the calculations. But tell me a story about the actual experience and it instantly jells with me, and why I love taking my projects out into the field.
.
So here's one of mine-
.
.
.
Not long ago we were testing some microwave equipment from the top of Pike's Peak - 14,115 feet, about 8000 feet above Colorado Springs and the Great Plains to the east. Calculate that line-of-sight horizon ....... :)
.
We sited the other end of the project as far out on those eastern plains - (its always fun to pull up in government vehicles, set up some weird equipment in the towne park, and get the looks we get from the residents..... :) ) -- as far out on the plains as our numbers said would work. And Bingo!... perfect pathways from perfect math.
.
Now confident, a second site, much closer in to us was setup- and-- Nothing.
........... In the clear, closer- conditions the same-- and it might as well have been site'd in a black hole. The 'calculations' said it should be anything but.
.
There were, of course, the variables- but they were supposedly taken into account when the calculations were done. Never-the-less, the project flopped at that point. I am still pondering what we missed.
.
.
Just anecdotal evidence that :
.
..................“The best laid schemes o' mice an' men, Gang aft a-gley.”
.
Or my favorite ;
.
...................."there are Pixies in those formulae."
.
.
..........................CF
 
Last edited:

Project25_MASTR

Millennial Graying OBT Guy
Joined
Jun 16, 2013
Messages
4,164
Location
Texas
I was going to add some more technical stuff to the discussion, but I think you-all have cover'd it marvelously.
_________________________________
.
I love anecdotes- for there is only so much I can glean from reading formulaes and doing the calculations. But tell me a story about the actual experience and it instantly jells with me, and why I love taking my projects out into the field.
.
So here's one of mine-
.
.
.
Not long ago we were testing some microwave equipment from the top of Pike's Peak - 14,115 feet, about 8000 feet above Colorado Springs and the Great Plains to the east. Calculate that line-of-sight horizon ....... :)
.
We sited the other end of the project as far out on those eastern plains - (its always fun to pull up in government vehicle, set up some weird equipment in the towne park, and get the looks we get from the residents..... :) ) -- as far out on the plains as our numbers said would work. And Bingo!... perfect pathways from perfect math.
.
Now confident, a second site, much closer in to us was setup- and-- Nothing.
........... In the clear, closer- conditions the same-- and it might as well have been site'd in a black hole. The 'calculations' said it should be anything but.
.
There were, of course, the variables- but they were supposedly taken into account when the calculations were done. Never-the-less, the project flopped at that point. I am still pondering what we missed.
.
.
Just anecdotal evidence that :
.
..................“The best laid schemes o' mice an' men, Gang aft a-gley.”
.
Or my favorite ;
.
...................."there are Pixies in those formulae."
.
.
..........................CF



If I ever get the time to do it, I've been wanting to test a PTP link across the Wet Mountain Valley. Thinking about a shot to Pikes while I'm at it…just for fun.


Sent from my iPhone using Tapatalk
 

quad_track

Member
Joined
Sep 13, 2017
Messages
66
First, the Longley-Rice model can be quite accurate, but there's also ITWOM for those edge cases. To accurately reflect environment conditions, you need to add land coverage data. I haven't followed the developement of Radio Mobile, but I assume it can make use of landcover data. Here in Europe, we have the Corine program which provides raster with a resolution of 100 meters.
Next, you want to make use of OpenStreetMap buildings, which come as shapefiles that can be imported as 3D models into your rendering engine of choice. I use Flightgear but plan on switching to OsgEarth as it's more lean and mean. With the 3D data, you can do two-pass ray tracing to account for reflections and diffraction in an urban environment. The big boys sometimes use Lidar data, and OsgEarth can also use that.
 

paulears

Member
Joined
Oct 14, 2015
Messages
769
Location
Lowestoft - UK
My own feelings are that until they can better estimate the quantity of rebar in concrete, the low level stuff will remain very hit and miss. One of my relatives is involved with new UK emergency service changeover, and they cannot get accurate signal strength predictions to work out the service area. Even worse, our Government couldn't keep the database of where the cell transmitter towers are and what frequencies and powers they have, so abandoned the centralised record keeping, and now trust the cell companies to keep their own records accurate and make them available on request.
 

quad_track

Member
Joined
Sep 13, 2017
Messages
66
It really depends on how much effort are you willing to put into it. For wide area coverage, if you have sufficiently accurate data, the statistical mode aka Longley-Rice will work well enough. For better results, you import a 3D model of your area and you run multipass ray casting on the GPU. Has pretty good results for multipath. It's a pretty computing intensive task, that's why it's done on the GPU. There are several papers about it somewhere on IEEE.
 

zz0468

QRT
Banned
Joined
Feb 6, 2007
Messages
6,034
It really depends on how much effort are you willing to put into it.

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?

For wide area coverage, if you have sufficiently accurate data, the statistical mode aka Longley-Rice will work well enough.

I've found Longley-Rice works quite well in suburban environments, especially when land use data is applied. Not so much in open rural environments, it has a tendency to be quite pessimistic in it's coverage predictions.
 
Status
Not open for further replies.
Top