Comparative Performance of Neural Network and Wavelet Based for Path Loss Prediction on Global System for Mobile Communication (GSM) in an Urban Environment

In this paper, GSM signal strength in the Dnepropetrovsk city was measured in order to predict path loss in study area using nonlinear autoregressive neural network prediction (NARNN) and the wavelet based one dimensional multilevel de-noising technique (1D MDT). In addition, a neural network clustering was used to determine the average GSM signal strength received in the study area. The two methods predicted that the GSM signal is attenuated in the study area with the mean square error (MSE) of 3.3978dB for the NARNN and 3.428dB for the 1D MDT respectively. NARNN demonstrated better prediction performance of 3.02 % than the 1DMT. We used these MSE values and modified the reference models. The neural network clustering revealed that -75dB to -95dB is received more frequently. This means that the signal strength received in the study is mostly weak signal.