Channel Estimation And Robust Precoding Transmission For Massive MIMO Systems

With the rapid growth of intelligent mobile terminals and the development of Internet of Things(IoT)in recent years,the demand for wireless data transmission continues to grow exponentially.Furthermore,the next generation of mobile communication systems have to provide users with high-speed,high-reliability,low-latency services.Massive multi-input-multi-output(MIMO)technology,which provides higher transmission rate and spectral efficiency,has become one of key technologies for the next generation of mobile communication systems.Motivated by these reasons,we investigate the channel estimation method in massive MIMO systems with low pilot overhead and study the robust precoding scheme according to imperfect channel information.First,the traditional channel estimation methods are studied,including least squares(LS),minimum mean-square-error(MMSE)and discrete cosine transform(DCT)based channel estimation algorithms.Theoretical results of the mean square error performance of these algorithms are then analyzed.The simulation results show that MMSE algorithm has the best performance.On the other hand,the traditional precoding algorithms,including matched filter(MF),zero-forcing(ZF)and regularized zero forcing(RZF)precoders are analyzed,so as their signal-to-interference-plus-noise ratio(SINR)performance.Finally,simulations of these algorithms show that these traditional precoding algorithms have significant performance loss under imperfect channel information.Second,for the unaffordable pilot overhead problem of traditional channel estimation methods in massive MIMO systems,a compressed sensing(CS)channel estimation method is proposed.Based on the jointly correlated Rician fading channel model,the sparsity of channel matrix in beam-delay domain is analyzed.Then,over-complete matrices are utilized to mitigate the energy leakage of channel in delay domain and increase the sparsity of channel in beam domain.Based on the modified joint correlation channel model and the sparsity of channel,we propose the modified orthogonal matching pursuit(MOMP)algorithm and energy-concentration based channel estimation(ECCE)method to estimate the support set of channel vector.The simulation results show that the ECCE method provides performance gains over the existing channel estimation methods and reduces the pilot overhead.Then,for the performance degradation of traditional precoder brought by imperfect channel information in massive MIMO systems,we proposed a robust precoder design scheme that maximizes the weighted ergodic sum-rate performance under imperfect channel state information.Based on the joint correlated Rician fading channel model,the priori model and posterior model of channel are analyzed,and the compressive sensing method of statistical channel state information acquisition is investigated.Then we utilize the Lagrangian multiplier and KKT(Karush-Kuhn-Tucker)conditions to solve the precoder design problem,and derive a iterative algorithm.As for the computational complexity introduced by the inverse of matrix in iteration,a low-complexity robust precoder is then proposed,whose complexity is significantly reduced.The simulation results show the performance of the low-complexity robust precoder is slightly worse than the optimal robust precoder.Furthermore,imperfect channel information has less impact on the performance of robust precoder,which is significantly better than RZF precoder.Finally,for the complexity problem of optimal robust precoder,another robust precoder design scheme is proposed,which maximize the signal-to-leakage-plus-noise ratio(SLNR)of each user.Based on the modified joint correlated channel model,SLNR is used as a criterion to not only maximize the power of each user but also minimize its power leakage on other users.Two different SLNR maximization robust precoding methods are proposed by utilizing Rayleigh-Ritz quotient result and generalized eigenvalue theory.Numerical simulations show that the multi-stream robust precoder has better sum-rate performance.The ergodic sumrate performance of the proposed robust precoder is much better than RZF precoder,and the imperfect channel information has less influence on the SLNR-based robust precoder.