Radio-wave propagation prediction model tuning of land cover effects

Abstract

The effect of land cover is incorporated in the radio propagation prediction algorithm of Q-Rap. It is implemented by optimizing both the effective height of the land cover, hence affecting obstruction-loss calculations, and by adding terms to the basic transmission loss algorithm. A complete set of separate coefficients to these terms is determined for each land cover type. The optimization method improves the standard deviation of the error from 9.6 to 6.3 dB for measurements and predictions done at 390 MHz. This is an improvement of 3.3 dB over the original model that comprises the free-space loss equation with obstruction loss calculations for multiple knife edges. At this frequency, the correlation coefficient between the measured and predicted values improved from 79.5% to 85.6%. At 2145 MHz, the optimization method improves the standard deviation of the error from 16.2 to 8.6 dB, as well as the correlation coefficient between the measurements and predicted values from 56.2% to 70.5%. The use of the correlation coefficient between the measured and predicted signal values, in addition to the standard deviation of the error and mean error as criteria to be used when evaluating propagation prediction models, is also proposed in this paper. A basis for best practices in tuning propagation prediction algorithms in radio planning tools using semi-empirical models is presented.