Updated: Feb 17
Comparison of field-strength measurements with simulation results derived from coverage prediction of Longley-Rice, ITU-R P.1546, and ITU-R P.1812 propagation models.
In the RadioPlanner 2.1_220216 update, it became possible to compare coverage prediction results with field-strength measurement data for TV and radio broadcasting projects.
While testing this feature, we made field strength measurements for several analog FM broadcast transmitters located on the Novosibirsk RTRS tower. Thus, we have an excellent opportunity to compare the measured signal levels with the prediction results using three different methods that RadioPlanner 2.1 allows: ITU-R P.1812, ITU-R P.1546, and the Longley-Rice method. And also compare the attenuation in various clutter types obtained experimentally with the attenuation determined according to the Rec. ITU-R P.1812-5, depending on the clutter height.
Antenna tower photo:
The measurement equipment consists of a Signal Hound USB-SA44B spectrum analyzer/measuring receiver, USB GPS Receiver, OPEK VH-1210 2dBi antenna (tuned to 106 MHz), laptop with Spike version 3.5.21 software. During the tests, the receiver antenna was placed on the car's roof (the phase center of the antenna is approximately 1.9m). Measuring the signal strength of FM radio using a vehicle-mounted measuring setup, in addition to the convenience and speed of such measurements compared to measurements with a fixed receiver, correlates better with the actual use of the broadcasting system because most people listen to the radio in the car.
Placing the antenna at the height of the car roof during measurements gives a correct result for assessing the coverage area than a fixed reception, in which it is recommended to place the antenna at the height of 10m.
The measurement equipment:
The measurement results (received channel power with point coordinates) were recorded in the Spike software with a 10-meter space. The measurement points number for different tracks was from 3500 to 4500.
In parallel with the measurements, an audio control of the radio station was carried out on a standard car FM receiver to compare the subjective reception quality with the measured signal level:
Measurement tests were carried out from the center of Novosibirsk along two highways 50k-19 (direction to Novokuznetsk) and P-255 (direction to Kemerovo).
The measurement area is the center and suburbs of Novosibirsk (low-density development, mainly 5-30 meters high) and highways passing through forests of variable density. This is a hilly-flat area with absolute terrain heights from 90m to 350m.
We made test measurements for two transmitters on different routes. The measurement results were saved in Spike as CSV files. After removing all the extra data that Spike saves, simple CSV files were prepared: Received channel power; Latitude; Longitude.
The measurement file:
These CSV files have been uploaded to the transmitter parameters:
We then compared these measurement results and the predictions made of each of the propagation models provided in RadioPlanner 2.1.
The prediction was carried out with the exclusion of the location and time variability component (percentage of location and time = 50%). Model parameters on the screenshots:
The clutter loss was determined for the condition of eliminating the mean error for each of the propagation models. Tuned Clutter Loss is attenuation in the corresponding clutter type, which is calculated under the condition of a mean error between the measured and predicted values equal to zero.
Below are screenshots of the measured and predicted field strength levels of one of the measurement tracks for different propagation models.
ITU-R P.1812 propagation model:
ITU-R P.1546 propagation model:
Longley-Rice propagation model
You can see a better fit of the predicted signal values to the measurement points for the ITU-R P.1812 model compared to other propagation models.
Standard deviation comparison results
Tuned Clutter Loss comparison results
1. The main criterion for propagation model quality evaluating is the standard deviation of the measured and predicted values. The smaller this value, the more accurate the prediction model. The table "Standard deviation comparison results" shows that for clutters with a significant statistical sample (100 or more points), the standard deviation in the range:
ITU-R P.1812 5.4-7.2 dB
ITU-R P.1546 4.7-7.4 dB
Longley-Rice 4.7-10.1 dB
We can conclude that the ITU-R P.1812 propagation model is the most accurate of all the above models. Standard deviation within 5.4-7.2 dB is a perfect result, which shows the correctness of the primary ITU-R P.1812 propagation model used in RadioPlanner 2.1 and spatial data - elevation matrix and clutters.
2. For the forest and roads in the forest, as well as for the urban, suburbs, and roads in the urban and suburbs, the absolute clutter losses obtained as a result of measurements coincide with the corresponding losses calculated according to ITU-R P.1812 with an accuracy of 1-3 dB.
3. On an "open/rural" clutter, measurements show an additional attenuation of 11.4 dB compared to the attenuation given by obstacle heights, according to ITU-R P.1812. This can be explained by the fact that on the tracks on which the measurements were made, there are no open/rural as such, mainly - this is an area surrounded by forest on all sides or (in rare case) buildings. The attenuation of 11.4 dB is just the same as the loss on forest or suburban.
Our tests do not pretend to be complete and highly accurate. Still, they clearly show that the result of coverage predictions in RadioPlanner is quite consistent with the results obtained by measurements with the right choice of propagation model.