Channel Estimation For Massive MIMO Systems Via Compressive Sensing

Accurate channel state information plays an important role in improving massive MIMO systems performance.And channel estimation method based on pilot sequences is usually used to effectively obtain the channel state information of the uplink and downlink channels in actual communication systems.However,pilot overhead required by conventional channel estimation will be unaffordable,especially for FDD massive MIMO.Compressive sensing technology can significantly save pilot resources and OFDM technology can combat frequency selective fading and inter-symbol interference.Therefore,this thesis mainly focuses on pilot design and channel estimation in FDD massive MIMO OFDM systems with the framework of compressive sensing and OFDM to effectively reduce the pilot overhead and improve the channel estimation performance.Firstly,a pseudo-random pilot design algorithm based on the genetic algorithm is proposed to solve the problem of too much pilot overhead and too low spectrum efficiency in massive MIMO systems.The non-orthogonal pilot at the base station is proposed.In addition,with the objective to minimize the cross correlation of the measurement matrix,this thesis improve the genetic algorithm and introduce the crossover and mutation judgment mechanism and the inner loop and outer loop mechanism to obtain pilot sequence with the minimum cross correlation,so as to ensure that pilot symbols are deterministically distributed in the frequency band.Simulation results demonstrate that the pseudo-random pilot design algorithm can effectively reduce the pilot overhead and obtain accurate channel estimation performance.Secondly,a channel model by exploiting the spatial correlation and the time correlation of massive MIMO channels is established and an adaptive channel estimation algorithm based on the locally common support algorithm is proposed due to the low accuracy and the long computation time of channel estimation,and the sparsity level of the channel needs to be set in advance in the existing researches of massive MIMO channel estimation.The proposed algorithm estimates the sparse support set according to the principle of maximum correlation,followed by the processing of sparse support set pruning.In addition,the proposed algorithm uses the linear minimum mean square error algorithm for matrix estimation to reconstruct the original signal.Simulation results demonstrate that the adaptive channel estimation algorithm can obtain accurate channel state information adaptively without predicting the sparsity level of channel,and it is capable of achieving accurate channel estimation performance and short computation time.