Improvement of Outdoor Signal Strength Prediction in UHF Band by Artificial Neural Network

Starting from measurements performed at 1140 MHz in an urban environment (Rio de Janeiro, Brazil), the signal strength measured is compared with the usual predictions calculated using the methods of International Telecommunication Union-Radiocommunication (ITU-R) Rec. 526-11/Cascade Knife Edge, and ITU-R Rec. 526-12/Delta-Bullington. The first model presented better adjustment to the data and, by using artificial neural networks (ANNs), this paper deals with the signal coverage prediction in a mobile channel calculated by the ITU-R model, which is used to train the ANN in order to reduce the average deviations. This hybrid application (ITU-R/ANN) showed efficiency, resulting in higher accuracy for the signal strength prediction, and an average improvement of 8 dB was achieved.