Satellite communication is one of the most important communication methods that are widely used nowadays. There are three basic systems of signal transmission: sender, receiver and the satellite. Sender will transmit the signal to satellite while the receiver will receive the signal from the satellite. However, signals that carry information within the transmission path are not the same as signals in the receiver path. The received signal may attenuate through a loss in desired power. The attenuation may be caused by rain rate, humidity, temperature, and wind speed. In this project, an intelligent forecasting system which is capable of determining the prediction attenuation signal based on the previous attenuation signal is developed to solve this problem. The Multilayer Perceptron (MLP) network is proposed to be used in developing the intelligent forecasting system. The MLP network is trained using two learning algorithms; Levenberg-Marquardt (LM) algorithm and Bayesian Regularization (BR) algorithm. This system is validated by the Index of Coefficient, R2, one step ahead (OSA) and multi step ahead (MSA) prediction test. In this study, it is proven that BR algorithm is the better learning algorithm as compared to LM algorithm to perform the forecasting system. The BR algorithm produces the higher R2 value which is 0.48906. This study proved that the MLP network has high capability in forecasting the satellite signal attenuation.
PEMBINAAN SISTEM PERAMALAN PINTAR UNTUK PELEMAHAN ISYARAT SATELIT MENGGUNAKAN RANGKAIAN NEURAL_Tan Chia Ling_E3_2007_875002949_NI