Performance Enhancement Of OFDM Using Intelligent System

Abstract

Abstract: Transmission of high data
is rate in a mobile environment makes the channel highly hostile. To combat with this problem, many techniques were proposed and developed. Orthogonal frequency division multiplexing (OFDM) system is a technique to combat this adverse channel. In this work, a method to enhance the performance of channel estimation in Orthogonal Frequency Division Multiplexing (OFDM) proposed by using different types of Back-Propagation (BP) for learning the Artificial Neural Network (ANN) to minimize Bit Error Rate (BER) when transmitting data. The proposed method includes learning Feed Forward Neural Network (FNN) and Recurrent Neural Network (RNN) by Conjugate Gradient algorithm, Quasi-Newton algorithm and Bayesian regularization. The comparison among Conjugate Gradient algorithm, Quasi-Newton algorithm and Bayesian regularization depends on the Mean-Square Error (MSE) convergence and precision generated in the BER calculation. This work is software implemented with MATLAB (R2013a) technical programming language.