For some time I have been looking for a good model or basis for estimating path loss of radio signals over water. This article presents two possible models that could be useful in predicting path loss over water at VHF frequencies.
I found some data about VHF radio propagation over water in the Federal Communication Commission (FCC) regulations for PART 80--STATIONS IN THE MARITIME SERVICE. The commission has provided some charts for the purpose of calculating the coverage of a shore station. A chart showing anticipated signal levels with over distance for various antenna heights allows deduction of path loss; the chart appears in Part 80.767 and is labeled "Propagation Curve."
The chart is very interesting because it shows signal loss with distance with several curves. There is a plotted line labeled "Free Space", which represents the theoretical best-case loss curve for a radio wave traveling a certain distance. There are also curves for real-world situations, with various antenna heights.
The curves show that for very short distance paths and very tall antennas, the path loss in the real world is the same as in free space, but once the path distance becomes more than a few miles, the characteristics of the path loss change remarkably.
I investigated this chart to see if I could deduce the path loss factor for the distance that was being used. I will describe my method. First, I looked closely at the free-space curve. I found the signal levels (dBu) for the distances (d) of 10 and 100 miles.
Miles dBu 10 83 100 63
When the distance changed by a factor of ten, the signal decreased by -20dB. This suggests the relationship is
dB = x log(d) -20 = x log(10) x = -20 / log(10) = -20
This is exactly what would be expected. The factor -20 log(d) is used in the free-space model of path loss.
Next, I investigated the other curves to see what sort of relationship between signal level and distance they represented. (Note that the x-axis in the chart is using a logarithmic scale.) Using the curve for an antenna of 100-feet, I used distance values of 5 and 50, where the slope of the curve appeared to be constant. I found the following values of signal and distance:
Miles dBu 5 75 50 19
When the distance increased by a factor of ten the signal decreased by -56 dB. This suggests the relationship of path loss and distance must be:
dB = x log(d) -56 = x log(10) x = -56 / 1 = -56
The curve on the chart seems to show that when when the distance between stations exceeds a few miles the path loss as a function of distance behaves according to the relationship dB = -56 log(d). This model describes a relationship between signal and distance where the path loss increases at a rate that is significantly more lossy than in free space. In another article, Estimating Path Loss on Marine Non-Line-of-Sight Paths, the path loss factor was found to be often cited in the range of -40 log(d) to perhaps -46 log(d), which is less than the -56 log(d) suggested by the FCC propagation chart.
The FCC's chart is part of a procedure to establish the coverage area of shore stations, so the intent may be to use a method that will result in a coverage area that is very reliable. This may explain why the propagation loss for distance is predicted to be much greater than the free space model would suggest.
In my earlier article (hyperlinked above) I estimated the path loss based on my estimate of the signal level received for a particular station, WXN69. I estimated the signal level received was -97dBm. This was just a rough estimate, based on the receiver having a rated sensitivity for a 10dB SINAD at a signal of -107 dBm, and judging (by ear and experience) the received signal to be perhaps 10dB stronger than that minimum signal level. If the signal level I observed were actually at the -107dBm level, then the path loss equation derived from that estimate would have needed a distance factor of -51.6 log(d) for (d) in miles. That value is not too far off from the FCC chart curve's behavior, which seems to suggest -56 log(d) is the appropriate model.
The FCC Part 80 chart makes clear that the free-space model estimate is not appropriate unless the path is very short and has a lot of ground clearance (to eliminate the influence of the ground and any reflections). It is also clear that any estimate of the effect of distance on path loss that suggests it will less the -40 log(d) is much too optimistic and probably completely unrealistic.
I am sure that at this point some readers—if there are any readers at all—are probably wondering what I am talking about. Let me explain a bit more in detail. When the signal loss or path loss for a certain distance is described as being
we are saying that the path loss increases with distance accord to that rate. Let us explore the increase in path loss when the path length doubles, as in a path of 10-miles increasing to a path of 20-miles. The path loss due to distance will be:
For 10-miles loss = 20 log(10) loss = 20dB For 20-miles loss = 20 log(20) loss = 26dB
When the path distance doubled, the path loss changed 6dB. Because there was decrease in signal, we say the path loss changed -6dB. The notation -6dB represent a ratio. In terms of a ratio of power, we can find the power ratio from the definition of dB for power, which is
dB = 10 log (P1/P2) where P1, P2 are two power levels
We can also evaluate that relationship to find the power ratio represented by the dB expression. That becomes
(P1/P2) = 10^(dB/10)
Now we evaluate for -6dB to find the ratio of powers:
(P1/P2) = 10^(-6/10) (P1/P2) = 10^-0.6 (P1/P2) = 0.25
In other words, when the path distance doubled, the signal power decreased to 0.25 of the original power, or one-fourth the original power. This is another way of describing the inverse square law relationship. I think many readers will be familiar with that behavior, but may have not realized that it was represented in the notation 20 log(d).
When models for propagation use a factor for distance of 20 log(d), they are assuming that the decrease in signals will just obey a simple inverse square law relationship. This is the behavior that would occur in free space, an ideal environment in which nothing else affects the radio wave as it propagates. In the real world there are other factors in the environment that affect the signal loss in propagation, causing the real-world signal decrease to be greater. This behavior is reflected in having a path loss model in which the distance factor is greater than 20 log(d).
More Investigating into path loss models permitted me to come across a reference to the c.1957 work of Egli, who proposed that path loss could be calculated by the relationship of:
Pr = 0.66*Gb*Gm * [(Hb*Hm)^2/d^4] *[40/f]^2 *Pt where Pr = received power Pt = transmit power Gm = gain of mobile antenna (not in dB) Gb = gain of base antenna (not in dB) Hm = height of mobile antenna (meters) Hb = height of base antenna (meters) d = path distance (meters) f = Mhz
This model contains several significant observations:
For more on Egli's model, and propagation path loss models in general, see Radio Propagation Models.
Egli's model is not very optimistic about a typical boat-to-boat path. Using the following input:
Antenna gain = 2 for both antennas, i.e. 3-dB Antenna height = 3-meters, i.e. about 10-feet, for both antennas Transmitter power = 25-watts Frequency = 156-MHz
the model calculates the received signal power as follow
Miles dBm 0.6 = -92.7 1.2 = -104.8 2.5 = -116.8 4.9 = -128.9
As you can see, when the distance doubles the path loss increases by 12dB, which is what is suggested by the 40 log(d) factor.
A really good receiver sensitivity is 0.25-microVolt, which is a signal level of -119dBm. Egli's model suggests that we will reach that signal level at a range of only 2.5-miles. The model does much better when antenna height is increased. For example, if one antenna is raised to 10-meters (33-feet), then the path distance for a -116.8dBm received signal doubles to about 5-miles.
The model is for propagation over smoothly rolling terrain with about 50-foot variation, so calm seawater will probably be provide a better path, that is, have lower path loss. The model is also statistical; it predicts a path loss that occurs about half the time.
I also found an on-line calculator that is said to use the Egli model. And I built my own spreadsheet version of an Egli's calculator. It shows very good agreement with the on-line above if the 0.66 factor is removed. That factor was probably to account for statistical probabilities.
Using the on-line calculator based on Egli's model (linked above), the path loss for the following situation
Antenna Height-1 = 100-feet (e.g., a shore station) Antenna Height-2 = 10-feet (e.g., a small boat) Frequency = 157-MHz Transmission distance = 30-miles
is calculated to be -160dB. We can now predict the signal level we might have for this path if we have the following characteristics in each station:
Transmitter power = 25-Watts Antenna gain = 3dB
A transmitter power of 25-Watts is a signal of +44dBm. Both the transmit antenna and the receive antenna have 3-dB gain. We can now predict the received signal level as follows:
TxPower + TxAntGain + RxAntGain = PathLoss = RxPower +44dBm + 3dB + 3dB + -160dB = -110dBm
A received signal level of -110dBm in a 50-Ohm system corresponds to a 0.7-microVolt of signal. Most modern VHF Marine Band radios have receivers with sensitivity of 0.5-microVolt for 10-dB SINAD, so the received signal on this path should be above the receiver's rated sensitivity.
For the path loss, transmitter power, antenna gains, and receiver sensitivity described above, the Egli's model predicts a useable signal at a range of 30-miles. Egli's model is based on gently rolling terrain over land with hills not more than 50-feet high. On an all-water path with no intervening terrain, the Egli's model should be a reasonably accurate predictor of radio coverage, especially over saltwater and calm seas.
As I noted above, the Egli's model includes an allowance for antenna height that is rather generous because every time the height is doubled the path loss decreases by 6-dB. Another way to say this is that doubling the antenna height causes the signal to increase by a factor of four. I inspected the FCC chart of path predicted coverage (above) and found that the same behavior was used in their predictive model: when the antenna height is doubled the predicted signal increases by 6-dB.
There seems to be some agreement in the two models on the influence of antenna height, and that influence seems really very generous. It is my understanding that both of these models are adaptations of real measurements of signal transmission over real terrain in the VHF and UHF spectrum, that is, they are not theoretical approaches but rather embodiments of actual observed phenomenon which have been then reduced to an empirical formula.
The influence of antenna height seems very significant in these two methods of predicting radio propagation, and this seems to suggest that antenna height really is very important, particularly when it is possible without too much mechanical difficulty to create a doubling of the antenna height, as would occur, say, if an antenna were raised to 10-feet from 5-feet in height. There is evidence in both the FCC data and Egli's data that such an antenna height increase will pay back a substantial dividend in increased signal (and thus increased range).
On this basis, it seems that one ought to prefer height to antenna gain in most instances. For example, it would be better to raise an antenna of 3-dB gain to double the height of an antenna of 6-dB gain. The increase in height is predicted to return a 6-dB increase in gain over most signal paths, while the increase in antenna gain would only return 3-dB (and then only in the main lobe which must be pointed at the distance station to achieve the gain).
If this article has raised any questions or if you have a comment, feel free to post it to a discussion thread reserved for that purpose.
Copyright © 2017 by James W. Hebert. Unauthorized reproduction prohibited.
This is a verified HTML 4.01 document served to you from continuousWave
Author: James W. Hebert
This article first appeared March 12, 2017. An earlier version appeared in a thread of articles in 2013