Research On Hybrid Precoding Technology In Millimeter Wave Massive Mimo Systems

To support the explosively growing mobile communication services,millimeter-wave(mmWave)massive multiple-input multiple-output(MIMO)technology can realize ultra-high rate and ultra-low delay transmission by utilizing abundant mm Wave spectrum resource and high multiplexing gain of large-scale antenna arrays.Mm Wave massive MIMO has become a key enabling technology of the next generation mobile communication system.Compared with traditional digital precoding,hybrid precoding is a critical signal processing technology in mm Wave massive MIMO systems.It can provide a compromise between hardware complexity and system performance.Therefore,hybrid precoding plays an important role for the development of mm Wave communications.However,mm Wave massive MIMO systems still face some problems,such as,nonideal phase shifters(PSs),frequency selective fading,inter-user interference,high computation complexity.To tackle these above problems,hybrid precoding technology is investigated to improve spectral efficiency of the mm Wave system.The contributions of this thesis are summarized as follows:First of all,hybrid precoding technology with low-resolution PSs is studied for single-user systems.To solve the nonideal PSs problem,a quantized scheme based on channel singular value decomposition(SVD)is proposed under any-bit resolution PSs(named SVD-Q).For SVD-Q scheme,assuming infinite resolution PSs,we derive the optimal phase values of analog precoder and combiner by SVD of channel matrix.Then,by using quantizer,we design analog precoder and combiner with any-bit resolution PSs.Furthermore,the digital precoder and combiner are calculated to maximize spectral efficiency.The results indicate that with the same resolution PSs,the proposed SVD-Q can achieve higher spectral efficiency than other schemes.In addition,an alternate scheme based on matrix rank-2 approximation(Alter-Rank2)is proposed to reduce the error of quantizer under one-bit resolution PSs.In this scheme,we propose a rank-2 approximation method to construct a binary codebook.According to this codebook,the optimal analog precoder and combiner are designed.Simulation results show that Alter-Rank2 can offer higher spectral and energy efficiency compared with the existing schemes.Next,OFDM-based hybrid precoding technology is investigated for single-user wideband systems.To design the shared analog precoding matrix caused by frequency-selective fading,a column-iteration scheme based on Jensen's inequality(JC-Iteration)is proposed.For JC-Iteration scheme,we first introduce a virtual variable assuming that analog precoder is fixed.Then,we derive the close-form solution of digital precoding matrix by maximizing spectral efficiency.Moreover,using Jensen's inequality,we transform the analog precoding design problem for frequency-selective fading into an analog precoding design problem for flat-fading.Meanwhile,the local optimal solution of this problem is computed by a column-iteration scheme.After that,we use the Sherman-Morrison formula to reduce the complexity of the column-iteration algorithm.In addition,an alternative minimization scheme based on phase matching(PM-AltMin)is proposed to further reduce the complexity of hybrid precoding design.In this scheme,the hybrid precoding problem is decomposed into a series of sub-problems.For each sub-problem,the optimal analog and digital precoding vectors are found by phase matching and least square methods alternatively,until obtaining the complete analog and digital precoding matrices.Simulation results show that both PM-AltMin and JC-Iteration schemes can achieve satisfactory performance,and JC-Iteration's performance is closer to the upper bound.Finally,hybrid precoding technology for multi-user systems is studied.To eliminate inter-user interference and improve sum-rate,mutual information criterion based two-stage scheme(MI-TS)is proposed.For MI-TS scheme,we first introduce the equivalent baseband channel capacity to design analog stage.In particular,the analog precoder and combiner problems are solved by a matrix-partitioning method.Then,in digital stage,the digital precoder and combiner are obtained by block diagonalization(BD)algorithm.The results indicate that compared with the existing hybrid precoding scheme,MI-TS can achieve higher multi-user sum-rate.Note that BD algorithm can eliminate inter-user interference completely,but it ignores the noise resulting in poor performance at low SNR.For this purpose,we propose a signal-to-leakage-and-Noise Ratio criterion based two-stage scheme(SLNR-TS).In analog stage,the analog precoder and combiner are designed to maximize the equivalent baseband channel gain.In digital stage,we derive the optimal digital precoder and combiner by SLNR criterion and Karush-Kuhn-Tucker(KKT)conditions.Simulation results show that SLNR-TS can make up the shortage of existing schemes,and achieve better performance.