DWT based Multi Carrier CDMA System using ANN-Aided Channel Estimation

Multi-Carrier (MC) Code Division Multiple Access (MC-CDMA) systems result from the combination of Orthogonal Frequency Division Multiplexing (OFDM) and Code Division Multiple Access (CDMA). These have become options for high data rate wireless systems. Due to the stochastic nature of the wireless channels, considerable amount of challenges still exist. Though traditional statistical techniques are available for such cases, learning based tools like Artificial Nueral Network (ANN) are also options. In this paper, ANN is used for channel estimation based on Levenberg-Marquardt (LM) training algorithm in MC-CDMA systems for multi antenna set-ups over different channel models. ANNs are preferred considering the fact that ANNs can better use the channel state information. An ANN in feedforward mode is trained using different pilot carriers with differing sizes under multiple channel states. The complexity of the proposed system reduces by using this approach and there is no requirement of any matrix computation. Further, the work achieves performance enhancement by replacing the traditional fast fourier transform (FFT) with discrete wavelet transform (DWT) in OFDM.