Pilot Optimization For Massive MIMO System Based On Genetic Algorithm

In modern wireless communications,with the increase of the freedoms of massive multiple-input multiple-output(MIMO)systems,the gains of diversity and multiplexing by multiple antennas can significantly improve spectral efficiency and energy efficiency.The channel state information(CSI)should be known to obtain the gains of massive MIMO.Conventional approaches for estimating CSI are the least square method and the minimum mean square error method employing the auxiliary pilot symbols.Conventional approaches of channel estimation cannot deal with the estimation of massive MIMO channles because of huge pilots overhead.In a massive MIMO system,with the increase of the number of transmit antennas,limited local scatterers bring about the channel matrices tending to be sparse.By employing the compressive sensing(CS)theory,the problem of channel estimation in massive MIMO system can be transformed into a compressed sensing reconstruction problem,and the problem can be solved by CS reconstruction algorithms.In the channel estimation model based on CS,the recovery matrix in CS depends on the vaule and the position of the pilots.So,we can use the irrelevance of the recovery matrix in the CS theory to perform pilot optimization.This paper focuses on the pilot optimization for downlink channel estimation in massive MIMO systems based on genetic algorithm(GA).The pilot optimization can improve the performance of channel estimation for the cases of single receive antenna and multi-antennas.By employing the criteria of design the recovery matrix,two kinds of pilot optimization methods are proposed:(1)Pilot optimization method based on partial Fourier matrix: several rows of Fourier matrices are selected as pilot symbols by GA,which makes the value of mutual-coherence of the recovery matrix be as small as possible.The value of mutual coherence of the recovery matrix is the maximum absolute value among the non-diagonal elements of its Gram matrix.Simulation results show that,by decreasing the value of mutual coherence of the recovery matrix,the error of channel estimation can be reduced 2-3dB effectively;(2)The optimization criterion is modified to minimizing the sum of the absolute values of off-diagonal elements of the corresponding Gram matrix of the recovery matrix.Simulation results show that,the improved optimization criterion can further reduce the channel estimation error about 1-2dB.In addition,when the terminal has multiple receiving antennas,the channel estimation can be modeled as a joint sparse signal reconstruction problem and be solved by employing the DCS-SOMP(Distributed Compressed Sensing Simultaneous Orthogonal Matching Pursuit)algorithm in Distributed Compressed Sensing(DCS)theory.We also research the issue of pilot optimization in this scenario.Simulation results show that the proposed pilot optimization method can also improve the performance of channel estimation in the case of multiple receive antennas.Compared with the channel estimation based on CS in the case of single receive antenna,the performance of channel estimation based on DCS in the case of multi-receive antennas is better.