Optimized GFDM Transceiver System Based on GA and ANN

Generalized frequency division multiplexing (GFDM) is a multiple access scheme promising for the 5G standard. It presents exciting features such as low out-of-band and low peak-to-average power ratio. This paper optimizes the GFDM transceiver system using a Genetic Algorithm (GA) and Artificial Neural Network (ANN). GA optimizes the pulse shaping filter, which gives a new coefficient based on optimization processes. The ANN presented as an enhancement estimation method compared with Least Square (LS), the data set used to train the ANN generated by the transmitted pilot, received pilot and channel response. This data was used to train the ANN and replace the LS part. The ANN by three methods to choose the best performance under the different values of neurons. The proposed system presents enhancement in BER by 0.011 at 20 dB and 30 dB.