Enhancement of BER of LTE System by Using DCT-Neural Network in Different Channel Models

Abstract

In this work the discrete cosine transform is proposed for LTE systems with the aid of feed-forward neural network as a suitable equalizer to retrieve the effect of channel within Rayleigh Faded channels. This system was implemented using the Quadrature-Phase Shift Keying as a modulation technique, and using different Maximum Doppler Shift, which represents the highest Doppler shift that can occur between the transmitter and the receiver in a given wireless channel. by using DCT-FFNN with different MDS values effectively mitigates signal distortion resulting from multipath propagation and common issues in wireless communication networks and demonstrates higher accuracy in predicting BER values. According to the research, the performance will be better at MDS 50 when compared to the rest of the MDS used in the paper. These advantages come with minimal loss in data rate and bandwidth and no additional expense in terms of power. The simulation results indicate that an FFNNs-based channel estimator outperforms the pilot-based channel estimator in LTE systems operating over a Rayleigh fading channel, because FFNNs have low complexity and can quickly and accurately adjust the signal strength of incoming signals based on their input. All LTE system models were implemented using MATLAB 2016.