Channel Estimation And Precoding For Massive MIMO Systems

Massive multiple-in multiple-output(MIMO)which employs a large-scale antenna array at the base sta-tion(BS)can deeply excavate spatial resources,thereby significantly improving the spectral efficiency and energy efficiency.The key point to improve the performance of a massive MIMO system is to obtain chan-nel state information(CSI)through channel estimation based on pilot signals.However,the pilot overhead consumes a large number of time-frequency resources.Therefore,it is vital to optimizing pilot allocation.In addition,the rich bandwidth resources at millimeter-wave(mm Wave)frequencies are essential for increasing the capacity of the 5th Generation Mobile Communication Systems(5G)and reducing communication de-lay.The precoding technology of massive MIMO can focus the signal energy to the dedicated user,which makes up for the propagation loss of mm Wave signals.Therefore,we aim for investigating channel estimation and precoding in massive MIMO systems,which mainly include classical massive MIMO systems for basic coverage and mm Wave distributed massive MIMO systems for hotspot coverage.The main innovations and contents of the thesis are as follows:Firstly,we propose joint angle-delay subspace based pilot reuse for broadband massive MIMO sys-tems which employ orthogonal frequency division multiplexing(OFDM)modulation.By investigating the spatially correlated massive MIMO-OFDM channels,we introduce a new concept of the joint angle-delay subspace.We reveal a relationship between the channel correlation matrix in the spatial-frequency domain and the joint angle-delay subspace matrix,and further apply a low-rank adaptive subspace tracking algorithm to estimate the joint angle-delay subspace matrix.Then,we propose a robust minimum mean square error(MMSE)estimator,and show that when the joint angle-delay subspaces of different users reusing the same pilot are non-overlapping,the MSE of the MMSE channel estimator reaches its minimum value.According to this optimality condition,we develop a low-complexity pilot scheduling method based on channel statis-tics.The simulation results show that the performance of the proposed pilot scheduling algorithm exceeds the traditional greedy pilot scheduling algorithm and the random pilot scheduling algorithm,and the proposed robust MMSE channel estimator can almost approach the same error of the MMSE channel estimator.Secondly,we propose channel estimation and hybrid precoding algorithms for distributed phased arrays based MIMO(DPA-MIMO)wireless communications.According to the distributed layout of the sub-arrays in the DPA-MIMO system,we propose a joint sparse channel model in the beam domain.In order to reduce the pilot beam overhead,we derive a deterministic beam training process of multi-sub-array cooperation.Based on this training process,we model the DPA-MIMO channel estimation problem as a structured sparse signal recovery problem in the beam domain.By taking advantage of the joint sparsity of the channel vectors in the beam domain,we propose two customized channel estimation algorithms,that is,the joint orthogonal matching pursuit(JOMP)algorithm and the joint sparse Bayesian learning-l₂(JSBL-l₂)algorithm.During the data transmission,we develop a hybrid precoding method based on sub-array grouping and serial inter-ference cancellation(SIC)to maximize spectral efficiency.The simulation results show that the proposed channel estimation algorithms can make full use of the channel characteristics to improve the estimation per-formance,and the proposed hybrid precoding algorithm can achieve better transmission performance with lower complexity for the sub-array structure.Thirdly,we jointly optimize downlink hybrid precoding design,uplink hybrid combining design,and uplink power control for the network-assisted full-duplex(NAFD)mm Wave DPA wireless transmission.We establish the transmission model in NAFD mm Wave DPA systems and formulate a realistic problem of max-imizing the two-way sum rate by taking into account practical constraints containing the power budgets for each phased array and user,the quality-of-service(Qo S)requirements of both uplink and downlink users,and the constant-amplitude property of phase shifters.To tackle this challenging problem,we equivalently transform the original problem into a minimization problem of both uplink and downlink weighted MSE and further propose a two-layer iterative algorithm based on penalty dual decomposition(PDD).In the inner iter-ation,we introduce a penalty convex-concave procedure(PCCP)to deal with the non-convex Qo S constraints and to avoid complicated initial point searching.We then provide a detailed computation complexity analysis of the proposed algorithm.The simulation results validate the effectiveness of the proposed algorithm and show the performance advantage of the NAFD mode over both the co-time co-frequency full-duplex(CCFD)and time division duplexing(TDD)modes in mm Wave DPA systems.Finally,we propose a hybrid precoding algorithm for mm Wave distributed antenna systems(DAS)that employ OFDM modulation.We model the wireless transmission including the delay spread differences in mm Wave DAS and formulate a realistic problem of maximizing the downlink sum-rate by taking into account the transmitting power constraint of each remote antenna unit(RAU)and the constant modulus constraint of phase shifters.To address this challenging problem,we first equivalently transform the original problem into a minimization problem of the downlink weighted MSE.Based on the PDD method,we propose a multi-RAU cooperative hybrid precoding algorithm.Meanwhile,in order to verify the effectiveness of the proposed algorithm,we provide a fully-digital precoding algorithm as a benchmark.The simulation results illustrate the performance of the proposed algorithm can almost approach that of the fully-digital precoding algorithm and reveal the impact of the delay spread difference on mm Wave OFDM DAS.