Channel Estimation And Low-Papr Precoding In Millimeter-Wave Communication Systems

After the world-wide popularization of the 4th generation mobile communication system, the design of the next generation mobile communication system confronts the problem of limited spectrum resources. Millimeter-wave (mm-wave) communication technology enables us to utilize the vast spectrum between 30GHz and 300GHz, which fundamentally eliminates the obstacle to the further improvement of commu-nication rates. Thus it has become a promising technology for the 5th generation mobile communication system. In recent years, mm-wave communication technology has experienced rapid development, and many technical solutions have been presented to accelerate the combination between mm-wave and mobile com-munications. During this process, the researchers have reached agreement on the mm-wave communication framework that incorporates beamforming and large-scale antenna arrays. On the basis of this fact, this dis-sertation investigates channel estimation and peak-to-average power ratio (PAPR) reduction in mm-wave communication systems. Specifically, our main contributions are summarized as follows:We propose a novel mm-wave channel estimator considering the equipment of large-scale arrays and beamformers. We first build the ray-tracing mm-wave channel model, based on which we can solve the channel estimation problem by estimating path propagation parameters such as angles of departure, angles of arrival, and path gains. Then we use two-dimensional beamspace multiple signal classification (2-D BMU-SIC) and least-squares methods to estimate the path directions and gains respectively. Finally, we analyze the feasibility of the proposed mm-wave channel estimator. Especially we focus on analyzing the 2-D MU-SIC spectrum ambiguity. Unlike the element-space MUSIC, the beamspace MUSIC might be disabled by inappropriate beamformers. We show how such an ambiguity arises, and provide sufficient conditions on the beamformers under which the MUSIC spectrum has no ambiguity and the number of resolvable directions is maximized. Simulation results show that compared with beam training and sparse recovery methods, the proposed mm-wave channel estimator has remarkable performance gain.We propose a novel mm-wave channel estimator taking into account 1-bit analog-to-digital converters (ADCs). The decease of quantizing precision can not only compensate the power consumption caused by the significant increase of mm-wave communication bandwidth, but also helps to decrease the overhead on feedback links. However, the usage of 1-bit ADCs makes channel estimation a very challenging problem. We first introduces the mm-wave channel estimator based on 1-bit dithered quantization (DQ), whose basic idea is to mix the input signal with some random or known control signal before quantizing. After that, we propose a new mm-wave channel estimator using 1-bit level-triggered sampling (LTS). The LTS is a sampling technique whose sampling time is random rather than deterministic, and it naturally produces 1-bit output. Finally, we perform asymptotic analysis about the proposed channel estimator. Especially we prove that it is consistent, asymptotically unbiased, and asymptotically normal. Simulation results show that compared with the mm-wave channel estimator based on 1-bit DQ, the proposed one achieves remarkable performance gain.The mm-wave channel estimators mentioned above can be used to obtain channel state information (C-SI). Based on the assumption that CSI is known, we propose precoding schemes of low peak-to-average power ratio (PAPR) to increase the power efficiency of power amplifiers (PAs). The design of low PAPR precoding schemes is combined with the hybrid beamforming structure with antenna subarrays. Under the hybrid beam-forming structure, we first define the PAPR on each PA, and then form low PAPR optimization problems with different constraints. Especially we consider the per-antenna peak power constraint and the crosstalk constraint. The average PAPR upper bound obtained by these precoding schemes is derived, which indicates that more antenna subarrays lead to lower average PAPR. Finally we provide algorithms of low complexity to solve the proposed optimization problems. The key step of these algorithms is to solve a proximal problem. The optimality of our solution to that proximal problem is verified by Karush-Kuhn-Tucker conditions. The simulation results under power clipping show that, compared with the zero-forcing precoder, the proposed precoding shcemes can achieve lower PAPR and lower bit error rate at the same time.