Calibration of modelling with survey data
Extended tutorial on using survey data to calibrate modelling accuracy.
For maximum accuracy, you should calibrate against local measurements where available. The process will help you select good settings for your model and clutter profile.
Steps:
1) Prepare your data as a CSV file containing latitude,longitude,rssi and remove any weak measurements below -100dBm
2) Import the data and compare it against your coverage heatmaps
3) Change the propagation model to find an appropriate model with the lowest overall error. Deygout diffraction is the most accurate.
4) Adjust the clutter profile to tune it to match the environment. For example, in a wooded area you may need to make the trees denser.
5) Continue to test and adjust clutter settings produce an RMSE error below 8dB.
Tips:
If your receiver is a phone then 3dB of error is receiver measurement error and your goal is to achieve below 5dB of modelling error.
Use a special colour key with 1dB steps and covers the range of your data eg. -30 to -100dBm. This doesn't need to be the same as the colour key you prefer but is important to achieve accurate calibration.
LiDAR data is best for LOS calibration but you can achieve a sub 8dB NLOS alignment with DTM and tuned landcover - as demonstrated :p
Documentation: https://cloudrf.com/documentation/05_...