A Novel Precoding Design For Frequency-Selective Massive MIMO Channel

In this paper, we newly introduced precoding algorithm for massive multi-user Multiple Input Multiple Output (MU-MIMO) in frequency-selective Rayleigh fading which divided into a number of subbands with flat fading. The introduced algorithms use a Supervised Singular Value Decomposition (SupSVD) method instead of SVD at the precoding level. The introduced precoding method exploits the underlying structure of the channel matrix of the bad fading subbands using the good ones. The introduced precoding model has a better representation of the randomly varying data of the channel to reduce the bad fading and the noise in the system. The overall effect of the proposed method gives an accurate representation of the channel in ultra-dense network communication. The simulation results compare the introduced precoder with linear precoding and detection methods. The overall results show the success of improvement with a 3 dB in terms of bit error rate (BER).