You are here

Question about Radio Mobile link

20 posts / 0 new
Last post
km5l's picture
Question about Radio Mobile link

This is my first attempt at this. A ham a few miles away has expressed an interest in linking with M3s. I think I already know it's not going to work at 11 miles given our elevation, but I do want to complete the exercise so I know how this works - and can start looking at the alternatives with nearby office buildings.

So here goes, in radio mobile, I established two sites, no problem, with elevation and position. On the link side I'm not sure how to input the data. For the M3 spec sheet located here:

We note that the TX gain is about 20 dbm or so at the "MCS 15" data rate. So, don't know how to move/translate those numbers to to tool where it's asking for Tx power (Watts), Tx antenna gain, and Rx antenna gain. Any help there?  Also, what about Tx and Rx line loss, etc? Or maybe the numbers are all net for the unit?

Patrick KM5L

Tx power (Watts)  
Tx line loss (dB)  
Tx antenna gain (dBi)  
Rx antenna gain (dBi)  
Rx line loss (dB)  
Rx threshold (μV)  
Required reliability (%)

AE6XE's picture
Tx power:   The MIMO devices

Tx power:   The MIMO devices have 2 xmit/rec chains, one for each antenna.  Consequently, I usually divide the power in half (to model what is going out one of the polarities).   However note, in MCS0 to MCS7, the same data is transmitted on both antennas-polarities and the receiver combines the energy to decode the signal.  Find a dBm to Watt conversion table and plug in the #.  

line loss:  I usually do 0.5dB loss for both Rx/Tx. These are  ~6" cables (rocket) or internal to the device (Nanostation), so minimal loss and this is probably a conservative value.

Antenna gain:  can be found on the specs.   Note, There is a way to google around and find ubiquiti antenna data that can imported into radio mobile, for each specifiic antenna, then just select the antenna.

Rx threshold:   3.99uV = -95dBm is what I typically use, but adjust for the MCS rate you are modeling for. 

Required reliability:  I usually leave this default.  

There's so many variables, that the coverage map gives a ballpark idea of what locations will work (significant to the rolling hills and mountain folks).  But then one just has to try it to know what the local environment noise allows  and rates can be achieved.

K6AH's picture
Radio Mobile is not a perfect

Radio Mobile is not a perfect science and people use it in various ways.  This is how I use it.

I always model for MCS0 and then look to see what else I might get.  Here are the parameters I use:

Tx power (Watts) .25
Tx line loss (dB) .5
Tx antenna gain (dBi) varies
Rx antenna gain (dBi) varies
Rx line loss (dB) .5
Rx threshold (μV) 4
Required reliability (%) 

I use .25 watts Tx Power.  While it's spec'd at .5 watts, that get's split between the vertical and horizontal domains; I use .5 dB loss in both Rx and Tx paths lines due to the coax jumpers; and I use 4 μV receive threshold because that's approximately equivalent to the MCS0 sensitivity of -94 dB.  You didn't specify the antennas you plan to use... and the 26 dB Rocket Dish or 18 dB sector antenna are candidates.  Those values get plugged into their respective fields.

When Radio Mobile comes back with its link analysis, it will calculate a Fade Margin.  From, my personal experience (and the values I've used, above), I know I need about 15 dB of margin for a basic connection.  Anything above will only increase the MCS rate for me.  For example.  I see that MCS15 requires 19 dB more received signal (94-75); I also see that the Rocket transmit power is 5 dB lower at that same rate... so I need a total of 24 dB (19+5) more fade margin... 39 dB in total (24+15) to achieve that data rate.

39 dB of Fade Margin is a lot, and I don't often achieve that (or MCS15 for that matter) on a link.

I hope this helps as a start.  Your own experiences will likely change the way you use Radio Mobile.  I started by modeling and then going out and seeing what results I got on the modeled links.  Eventually I established this baseline and it's worked out pretty well for me.  I have found that the UBNT airLink tool to be less than helpful.

Andre, K6AH

K6AH's picture
Figuring out the MCS Rate

If you telnet into the node, the following command will indicate the MCS rate the device is running:

cat /sys/kernel/debug/ieee80211/phy0/netdev:wlan0/stations/*/rc_stats

Here's an example of a node I have in my backyard pointing to a regional AREDN backbone node around 26 miles away,  The Node Status screen indicates -73 / -95 / 22 dB S/N/R.

type         rate     throughput  ewma prob   this prob  retry   this succ/attempt   success    attempts
HT20/LGI     MCS0            5.6      100.0       100.0      1              0(  0)         1           1
HT20/LGI     MCS1           10.5      100.0       100.0      4              0(  0)         4           4
HT20/LGI     MCS2           14.8      100.0       100.0      5              0(  0)        93          93
HT20/LGI     MCS3           18.6       97.7       100.0      5              0(  0)      1380        1416
HT20/LGI  tP MCS4           25.1       99.9       100.0      5              0(  0)     31688       33264
HT20/LGI     MCS5            8.6       25.8       100.0      0              0(  0)       175        3495
HT20/LGI     MCS6            0.0        0.0         0.0      0              0(  0)         1        3495
HT20/LGI     MCS7            0.0        0.0         0.0      0              0(  0)         0        3495
HT20/LGI     MCS8           10.5      100.0       100.0      0              0(  0)         1           1
HT20/LGI     MCS9           18.6       99.9       100.0      5              0(  0)       368         380
HT20/LGI     MCS10          25.1       99.9       100.0      5              0(  0)     37921       38776
HT20/LGI T   MCS11          30.3       99.9       100.0      5              0(  0)    439091      448760
HT20/LGI     MCS12          14.1       33.2       100.0      6              0(  0)      4482        8447
HT20/LGI     MCS13           0.0        0.0         0.0      0              0(  0)         0        3495
HT20/LGI     MCS14           0.0        0.0         0.0      0              0(  0)         0        3496
HT20/LGI     MCS15           0.0        0.0         0.0      0              0(  0)         0        3495

The "T" in the 10th character position indicates the current MCS rate, and a "t" indicates the current fallback rate.  In this case the link is running MCS11 at 30.3 Mbps.

Note that the command returns this chart for each of its connections.

Andre, K6AH

km5l's picture

Ok, the result was pretty pitiful but I expected that. At least now I can move into some other link scenarios and see what happens.

Thanks to you both.

KE2N's picture
I think that I shall never see - RF attenuator as good as a tree

With apologies to Joyce Kilmer

Strictly-speaking it is not all attenuation - some of it is forward-scattering that messes up the coherence of wave fronts so that they do not add up properly at the receiver. So it looks like attenuation, but does not behave exactly like an attenuator.  One big tree can have more attenuation than you expect; a forest can have less attenuation that you might calculate. Leaves and twigs that are 1/2 wavelength long have the biggest impact.

The problem with RadioMobile is that it really underestimates the effect of passing through trees, or tree tops. It also does not know what your path really looks like in terms of ground cover and obstacles. It is probably the wrong tool for propagation through trees, but you can try to increase the "density" number in Options/Elevation data to match your field test results. First look at the point-to-point path (tools/radio link - view profile) and see what objects the land cover file thinks are in the path (you might be surprised). Then adjust the height and density of those objects (options/elevation data).  The highest density I think you can use is 1000%.  Someone in the RadioMobile user group tried to make trees match the actual measured 3 GHz results and needed a density number > 1000%. So that may not be the answer.

There is a hidden feature to use an alternative attenuation model in RadioMobile.  You need to edit the "landheight.dat" file. For example:

​ 0, 0
 15, 100
 25, 60
 15, 100
 15, 60
 15, 70
 10, 70
 5, 10
 1, 10
 1, 10
 1, 5
 1, 5
 0, 0
 10, 100
 15, 115

The first set of numbers comes from the GUI.  The last one is a line *you* add, saying to use model #2 with parameters 0.2/0.5/0.8 (in this example). 
This is the CCIR 236-2 model (Google is your friend).The parameters are k, x and y, respectively.
If you add this line, the numbers will appear in the GUI as shown in the attached JPG.
Note: this model then applies to ALL the objects in the land cover file, not just trees.

I would like to calibrate this model against field measurements of attenuation due to ground clutter (aka trees) versus height, but nobody has given me a quad copter yet ;-)

With the numbers shown (and not based on anything in particular) I see results that seem to me a lot closer to reality; i.e., only very short paths work in leafy areas. 

Image Attachments: 
An important reminder for

An important reminder for those using SRTM (RM ONLINE for sure and often wth the software version) "ground" can be defined as the TOP of the tree line (and top of buildings) depending upon how dense the canopy level is.

This becomes very important when entering the height above ground for nodes. If canopy is at 30ft and you enter 50ft you could be telling Radio
Mobile your node is 80ft Above Ground Level.

KE2N's picture
SRTM data

The elevation data itself was not seriously affected by tree canopy

"... the SRTM elevations for evergreen forests are, on average, nearly 2 meters above the ground surface elevations. Interestingly, the bias for deciduous forests is much lower at less than one-half meter. A partial explanation for this observation may be that SRTM data were collected in February, which represents leaf-off conditions for deciduous forests in the conterminous United States, and more of the radar signals probably penetrated the relatively open canopy and reflected from the ground surface."

In the PC version of RM the antenna height is clearly intended to be the height above ground (attached image)
and where the ground level is, I think, is shown rather unambiguously (attached image - yellow line).

Image Attachments: 
I'v got a repeater site where

I'v got a repeater site where IIRC ground level is actually 30foot (+/- .5m GPS accuracy) below SRTM data.

Mind you this is California we don't necessarily have seasons on our trees even for deciduous trees but your right I had forgotten about that factoid that many areas may have been "leafless" and may not be as serious as I've experienced. 

KE2N's picture
srtm errors

"ground truth" studies have shown that SRTM 30-meter data can be off 9 meters in absolute elevation level, or 7 meters in relative elevation (so-called 90% errors).
It's just not due to trees.
Like anything with a computer, you have to do some reality check on your data.

Another option

I use Radio Mobile lots as I plot repeater systems or amateur base station antenna patterns. With AREDN I have come to trust as a first test of potential link success. I like this free app as it has all the Ubiquiti equipment already built into its database and the mapping interface appears to be pretty darn accurate. Once I run this to verify the line of sight between links for given locations I will then go to Radio Mobile for the propagation study.

In you can also layer in a hybrid map showing buildings and streets where you can quickly drag and drop your pin to potential locations. Give it a go.


KE2N's picture

Agreed:  airLink is nice in some ways. Although reported signals are hopelessly optimistic, unless you have mountaintop-to-mountaintop topology. You can lower your link right into a forest and airLink will say everything is fine. But as long as you understand that, it is useful.

Another one I like is ""  You have to imagine the tree layer, but I like the way it interfaces with the googlemap/googleearth tools.  Google has taken to making 3-D maps of more populated areas and they even show individual large buildings.  Another nice feature is that you can click on a spot on the profile and it will take you right to the map and show you what is there.  You can save the URL they give you and recall it later quickly.  

The original use of it is to identify mountain peaks you can see from any given location - useful in areas that actually *have* mountains  ;-)


n1cz's picture
It is fundamental that 0.6

It is fundamental that 0.6 Fresnel clearance MUST clear tree tops. Not as black and white at 900 Mhz, but nearly so with 900 being slightly more forgiving. This clearance is a must for 2.4 GHz and up. I have made 100s of microwave paths. Ham radio can cut a lot of corners versus any equivalent commercial system design for lower frequencies and modes, but not when it comes to 2.4 GHz and above. Unless, you are ok with the frustration and links that work some of the year.

If there is a rise or a ridge in the terrain shown on the path profile that is within 20 meters (per airLink) of 0.6 Fresnel in the path, drive to the area and make a physical survey. APRS is very helpful in locating yourself with the areas of interest. Also, the Boy Scout Handbook taught us how to estimate tree heights ... although I never understood why they needed to know. There might be building. It might be a bare ridge and not a problem. Or, as I found once in PA, an enormous coal mine slag pile in the way that the field survey did not initially recognize!

I think airLink is a fantastic tool with instant results, but some help from the real world is necessary.

-Jeff W9NIZ


Regardless of what modeling tool is used, you never know for sure that the link will work until the equipment is in place! After the modeling, I usually end up with a pair of binoculars or high power rifle scope to check site-lines before hanging nodes.

KE2N's picture
LOS by looking

I put a web cam with a 20x zoom lens on the tower and it is my favorite "propagation tool".  Visibility around here is variable, having the camera permanently up there, gives me a chance to pick a day when the "seeing" is good and snap some photo's.  The camera displays the heading to a fraction of a degree so I can plot the sight line on Google maps. 

In the attached photo, the buildings just beyond the airport are 13 miles and the faint ones in the background are about 23 miles. Time-of-day (sun angle) makes a lot of difference in what you can see. But the farthest I have seen is about 30 miles. This is not Montana.

Image Attachments: 
AE6XE's picture
Ken,  what ipCam model # is

Ken,  what ipCam model # is this?    We have a similar setup here in SoCal,  although no heading display.   I've been matching up google earth shots and water tanks to figure out the heading :) .   

We're putting in equipment on lab-campus at JPL in Pasadena this Saturday.  Here's a validation of line of sight pic.  That's Pasadena with the building cluster as near as we can tell (if I'm pointing the right direction) from Pleasants Pk.  This is about 40 miles through LA smog to Pasadena, the campus is 42 miles.  I found a rare day where I could see this far.   I'm waiting for the perfect day to take pics and see how far away it's possible to see.


Image Attachments: 
KE2N's picture

Nice smog. Nice mountains too. 
The camera is a Canon VB-M40 which I think is obsolete now.  But there is a 41/42/43 as well.

N2MH's picture
LA Smog

As Johnny Carson used to say, "I don't trust any air I can't see!" :-)

Camera info


what at camera are you using? I am looking for a couple to put on my main repeater site in Redlands both for security and line of site. If I find something of good quality and price I will also be placing them at a local hospital as well as up at CSUSB.


AE6XE's picture
Model is  PT7B022S200   http:
Model is  PT7B022S200

We love the optics.  Just snapped these pics from video over 30+ miles RF to Pleasants PK.  Looking over LA sun down, see white dot on right side.  This is LA Doppler radar site.   See zoom in at low light, can count the hikers on weekends.

Image Attachments: 

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer