Evaluation of radio propagation parameters for field strength prediction using neural networks

Abstract

Radio propagation models for field strength prediction are essential for designing and installing a mobile radio communications system. Typical backpropagation neural network (BPN) models with different input parameters are developed and used to evaluate and assess the relative importance of a set of radio propagation parameters for field strength prediction. The NN models are trained on data measurements of propagation loss with terrain information taken in an urban area (Athens region) in the 900 MHz band. The performance of all NN models is evaluated by comparing their prediction error statistics of average value, standard deviation, root mean square and the correlation between their predicted values and the true data measurements. The NN model with the best performance provides an indication of the most important set of parameters for field strength prediction.