Predicting field strength with a neural ray-tracing model

A model of the field strength prediction for a mobile phone environment is presented. Ray tracing techniques are used to simulate the propagation of radio waves in a geometrical urban environment composed of terrain, buildings, and vegetation. The use of neural networks gives a correction of biases generated by the unknown properties of the environment and the algorithmic simplifications. We expect this model to overcome some of the limitations inherent in both empirical and theoretical models: respectively, the inability to adapt to non-classical situations, and exponential computation times. Comparisons with measurements give good results in several French cities with different frequencies