Modelling of propagation path loss of GSM–900MHz frequency band by artifical neural networks

Communication is constant and efficient for wireless communication with analyze radio propagation. Transmitter and receiver antennas can be optimized with developed various model that mesaure RSS level informations (received signal strength) for moving receiver system and wireless networks can be designed. Maximum communication range can find with propagation path loss paremeters. In this study, propagation path loss mesaurements is modeled by artificial neural networks for the base transceiver station that using GSM - 900 MHz bandwtih. The propagation path loss results are calculated with some empiricial propagation path loss models and these results compare with alternate path loss model that designed with some artificial neural networks. 361 received signal strength datas that has different distance are studied, then 5 input parameters and 1 output parameters are defined for artificial neural networks. Multilayee neural networks (MLNN), Radial basis function neural networks (RBNN or RBFNN) and Self organizing map (SOM) models from artificial neural networks varieties are developed with 3 fold mode. Okumura and Hata models from empricial propagation models are used for this study. Keywords: Gsm, electromagnetic wave, artificial neural networks, propagation path loss.