Research On Block Sparse Channel Estimation For 5G

Accurate channel estimation can effectively improve the spectrum utilization and energy efficiency of massive MIMO and three dimension(3D)massive MIMO systems,which becomes an indispensable key technology for designing next generation communication system.However,because of the number of antennas exponentially increase at base station(BS)in massive MIMO system and the channel dimension extending in 3D massive MIMO system,much more pilots should be inserted for channel estimation,which makes the spectral utilization substantially decrease.Based on above situation,how to make a balance between system performance improvement and pilot overhead decrease is a hot topic in signal processing area.Compressive sensing(CS)theory can make fully use of signal sparsity to obtain discrete samples randomly,which also means the pilot overhead will reduce while holding the reconstruct signal accuracy.Thus,CS has been widely used in channel estimation researchs.For the above massive MIMO and 3D massive MIMO systems,most current channel estimation techniques focused on either one-dimensional(1D)delay-spread domain or two-dimensional(2D)delay-doppler domain.There are few literatures considering the block sparsity of multi-dimensional antennas in angular domain.In order to fully utilize this pontenial channel structure characteristic,we proposes an effective block sparse algorithm and applied it into the channel estimation for massive MIMO and 3D massive MIMO systems respectively.What is more,we also proofed that the proposed algorithm can further enhance the spectrum utilization and energy efficiency.The main contributions are summarized as follows:Firstly,a novel block sparse channel estimation algorithm is proposed.We figured out that the traditional sparse channel estimation alogrithms have higher pilot overhead and lower signal reconstruction precision drawbacks.Then we fully exploit the block sparsity of wireless communication system and proposed an improved block partition sparse channel estimation algorithm.The simulation results show that the proposed algorithm can reduce the acquisition of measurement data while ensuring the reconstructed signal accuracy,and thus effectively reduce the pilot overhead.Secondly,considering the massive MIMO system,we applied the proposed algorithm to the scene which equipped with large numbers of antenas at base station.First of all,according to the spatial correlation of massive MIMO system,the channel block sparsity is fully exploited,and the channel estimation performance by using the improved algorithm also been analyzed.Besides,by considering a single-user massive MIMO system equipped with uniform linear array(ULA)antennas at BS,we make some matlab simulations,and the simulation results show that the error bits rate is lower and the normalized mean square error(NMSE)is smaller by utilze the novel channel estimation algorithm.Moreover,the reconstructed signal is very close to the original one.Thirdly,for the multi-user 3D Massive MIMO system channel estimation,we taken into account the antenna elevation angle at BS and make fully use of the freedom in spatial domain,establish a multi-user system model and proposed a novel pilot insert scheme.Firstly,based on the angle domain of 3D MIMO system,combined with the large-scale antenna technique,we create a multi-user 3D massive MIMO system model in time-angle domain.Secondly,in order to reduce the pilot overhead and enhance the estimation accuracy,a non-orthogonal pilot scheme is proposed.Finally,compared with traditional sparse channel estimation algorithms,the simulation results show that the proposed algorithm requires lower training sequences due to the elevation angle gain,and enhance the system spectrum utilization rate performance at the same time.