Decoding Method Based On Complex ICA For Multi-Cell Massive MIMO Uplink Systems

As one of possible key technologies for next generation wireless communications,massive Multiple-Input,Multiple-Output(MIMO)has attracted significant interest recently.The prominent feature of a massive MIMO system is that a Base Station(BS),equipped with a large number of antennas,can serve several single-antenna users simultaneously in the same frequency band.In fact,results of research have shown that,if the Channel State Information(CSI)is available for the receiver,the system can provide high data transmission rate and high energy efficiency with simple signal processing.But the performance will be degraded greatly if this CSI is not available.Usually,the CSI is acquired by a training system.That is,users send predesigned pilot signals,and the BS estimates the CSI when the pilot signals are received.To ensure accuracy of the CSI estimation,the pilot signals sent by different users should be designed as mutually orthogonal.But this orthogonality will consume a lot of resources when the BS communicates with several users in the same frequency band simultaneously.Moreover,in a multi-cell,massive MIMO system,co-channel cells will be set up at a close range,and hence,inter-cell interference will be inevitable.It is also possible that the different users send the same pilot signals at the same time,which causes severe damage on the CSI estimation and on the performance of the system.This phenomenon is called as pilot contamination.Pilot signal reuse in neighboring cells causes the pilot contamination,and performance will be degraded significantly in a massive MIMO system.In this paper,a new decoding method is proposed to alleviate the degradation.In the proposed method,the Principle Component Analysis(PCA)is used to reduce the dimension of the received signals from both inter-cell and intra-cell.Then,the complex Independent Component Analysis(ICA)is implemented to estimate the channels,the Mean Minimal Square Error(MMSE)decoder is employed to decode the transmitted signals,and few pilot signals are used to overcome the ambiguity caused by the complex ICA.At last,fourier transform based unitary space-time codes are used to optimize the pilot signals for a better performance of massive MIMO system.In our method,orthogonality or asymptotic orthogonality of the channels does not be required.Moreover,the path-loss and shadowing factors can be unknown.Simulation results show that the performance of our decoder is better than the MMSE channel estimation based MMSE decoder.Moreover,error floor does not appear even when Signal-to-Noise Ratio(SNR)is high,while it occurs if the MMSE decoder is used.