Research On Sinal Detection Algorithm In MIMO-GFDM System

Generalized frequency division multiplexing(GFDM)is one of the multi-carrier transmission schemes for 5G physical layer candidates.Compared with Orthogonal Frequency Division Multiplexing(OFDM),the GFDM has certain advantages in spectrum utilization,out-of-band(OOB)radiation,and peak-to-average power ratio(PAPR).Because GFDM can set the parameters that meet the requirements of different application scenarios,it is a very promising modulation waveform.Multiple Input Multiple Output(MIMO)technology is an indispensable part of future mobile communications.Any multi-carrier transmission scheme must be combined with MIMO.In this paper,according to the characteristics of MIMO-GFDM system,the algorithm of signal detection suitable for this system is selected and the algorithm is improved.The interference in the system is preprocessed.The details are as follows:1.This paper introduces the GFDM modulation principle and three linear demodulation methods.Then,Comparing the performance of GFDM and OFDM by simulation of matlab,and the factors affecting the BER of GFDM system are summarized.Then,the Alamouti space-time coding principle and the combiner principle are briefly introduced.The Alamouti space-time coding is combined with GFDM to form the STC-GFDM system and the TR-STC-GFDM system.The differences in bit error rate performance,spectrum utilization,and delay conditions between the STC-GFDM system and the TR-STC-GFDM system are compared.2.The characteristics of MIMO-GFDM system are studied.Aiming at the problem that the equivalent channel dimension of this system is very large,which leads to the complexity of the traditional signal detection algorithm in MIMO is too high,the Reactive Tabu Search(RTS)algorithm is applied to it.The simulation verifies that the MMSE-RTS algorithm can be applied in the MIMO-GFDM system and TR-STC-GFDM.The BER performance and computational complexity of MIMO-GFDM in ZF detection,MMSE detection,MMSE-SIC detection,QRD-M detection,and MMSE-RTS detection are compared.The simulation results show that the MMSE-RTS algorithm has lower bit error rate than the comparison algorithm,especially in low-order modulation.In terms of complexity,the MMSE-RTS detection is lower than the MMSE-SIC detection,and the two linear detection algorithms are in one series.3.Aiming at the high complexity of the initial value inversion process of MMSE-RTS algorithm,the inverse part of the matrix is subjected to Cholesky decomposition,and the initial value is iteratively calculated by Sherman-Morrison formula.Inverting a high-dimensional matrix into a sparse matrix iterative multiplication reduces the high computational complexity of the initial value inversion.4.According to the characteristics of the diagonal matrix formed by the inter-antenna interference matrix of the frequency domain MIMO-GFDM system,it is decomposed into a plurality of frequency domain MIMO channel matrices.Then frequency domain equalization is performed separately,which can completely remove inter-antenna interference and greatly reduce the complexity of frequency domain equalization.5.After MIMO-GFDM performs frequency domain equalization and matched filter,there will be system self-interference.It is proved that the self-interference matrix of the GFDM modulation matrix with two different structures after matched filter demodulation is composed of block cyclic sub-matrices,and the two schemes of Block Inverse Discrete Fourier Transform(BIDFT)precoding for MIMO-GFDM are derived.The simulation compares the bit error rate in MIMO-GFDM system with Hadamard precoding,DFT precoding,two BIDFT precoding,MF demodulation and ZF demodulation.It is proved that the bit error rate performance of BIDFT precoding is much better than hadamard precoding and DFT precoding.The effect of the number of subcarriers and the number of subsymbols on the bit error rate performance of the two BIDFT precoding schemes is investigated.The simulation compares the peak-to-average ratio of Hadamard precoding,DFT precoding,and two BIDFT precoding in GFDM system.