低精度量化通信系统接收技术研究

发布时间：2018-06-12 16:39

本文选题：功耗效率 + 低精度量化　；参考：《中国科学技术大学》2014年博士论文

【摘要】：在无线通信领域,人们一直致力于系统功耗效率的提高。无线系统的功耗主要包括信号的辐射功耗和处理功耗。随着空间复用率的提高,无线通信呈现出短距化的趋势,使得辐射功耗在整个系统功耗中所占的比重逐渐降低,处理功耗(尤其是接收端处理功耗)的优化成为提高系统功耗效率的有效手段。接收端处理功耗主要包括模数转换器(Analog-to-digital Conversion, ADC)功耗和基带处理功耗两部分,其中ADC的功耗与量化精度成指数关系,与采样频率成线性关系,而基带处理功耗与量化精度近似成平方线性关系。随着未来通信系统向着大带宽高速率方向发展,传统高速高精度ADC将因其巨大的功耗与随之产生的海量数据处理而成为实现低复杂度低功耗系统的重要制约因素。为此,低精度量化ADC被引入到数字接收机中,并被认为是一种能够有效解决大带宽高速率接收机复杂度和功耗的手段,受到学术界和工业界的广泛关注,具有重要的研究意义和实用价值。本文面向未来通信系统宽带化和功耗效率优化的迫切需求,研究了低精度量化下的接收技术,旨在为大宽带通信提供一系列低成本、低功耗的解决方案。本文从接收机设计与性能分析、系统非理想因素的影响、链路码间串扰的影响三个方面展开研究工作,主要的研究成果和贡献如下：针对接收机设计与性能分析,本文研究了采用正交调制的低精度量化通信系统中接收机的设计及其性能分析。本文首先给出了QPSK调制下的最优单比特接收机,并以失真度为指标简化了性能分析。本文进一步给出了次优的线性单比特接收机以降低实现复杂度和增加鲁棒性。仿真结果表明,相对于拥有完美信道状态信息(Channel State Information, CSI)的单比特匹配滤波器接收机,该接收机在加性高斯白噪声信道(Additive White Gaussian Noise, AWGN)中有约3dB的信噪比(Signal-to-Noise-Ratio, SNR)损失,而在密集多径信道中仅有约1dB的SNR损失。本文给出了Z-阶量化下的最优接收机,求得了比特对数似然比。通过比特对数似然比的高斯近似,给出了Z-阶量化最优接收机的误比特率(Bit Error Rate, BER)性能分析。定量给出了低精度量化接收机性能与量化精度的关系。理论与仿真结果表明,当量化阶数大于8时,继续增加量化阶数所获得的性能增益小于0.1dB。利用该方法对相位差的影响、量化门限的影响以及性能增益与量化精度的关系等的分析与仿真结果高度吻合,证明了该分析方法的有效性,能够给低精度量化接收机的实际应用提供设计指导。此外,本文还给出了基于最小均方误差(Minimum Mean Square Error, MMSE)的可实用线性分段接收机,以降低接收机实现复杂度和增加鲁棒性,并给出了相应的信道估计方法。针对非理想因素的影响,本文重点研究了IQ不平衡对QPSK单比特接收机的影响及相应的对抗策略。证明了幅度不平衡对于单比特接收机没有影响,而相位不平衡和收发相位差共同作用会严重恶化单比特接收机性能。提出了一种基于双训练序列的单比特接收机和一种加权双训练训练单比特接收机,能够在AWGN信道和稀疏多径信道中有效对抗IQ不平衡导致的性能损失,特别是在高SNR区域。对信道编码环境提出了发送端星座旋转的策略,通过该旋转策略提供的“相位分集”能够在AWGN信道和稀疏多径信道中为码率1/2～7/8的卷积码提供约O～3dB的SNR增益。提出了一种8-扇区相位量化接收机,考察了其对于旋转相位的鲁棒性,设计了相位量化下具体接收机实现,从而几乎完全消除了IQ不平衡导致的性能损失。由于可以通过多径提供相位分集,IQ不平衡对于密集多径信道影响较小。针对链路码间串扰的影响,研究了将低精度量化接收机应用于高速宽带系统的过程中遇到的链路码间串扰问题。利用高速宽带系统的信道特性,对码间串扰进行高斯近似。基于码间串扰的高斯近似,给出了低强度码间串扰下的单比特接收机设计,并对该接收机从输出信噪比损失和BER两个角度进行性能分析。理论和仿真结果表明在250Mbps及以下速率时,该接收机相比于没有码间串扰的情形其性能损失小于1dB,表明无需复杂均衡技术。针对高强度码间串扰,提出了一种基于判决反馈和迭代解调的单比特时域均衡算法,能够有效对抗码间串扰的不良影响。对于500Mbps的情形,该均衡算法相对于没有码间串扰的情形仅有约1.2dB的SNR损失。
[Abstract]:In the field of wireless communication, people have been devoted to improving the power efficiency of the system. The power consumption of the wireless system mainly includes the radiation power of the signal and the power processing power. With the increase of the spatial reuse rate, the wireless communication presents a tendency of short distance, which makes the proportion of radiation power consumed in the power consumption of the whole system gradually reduced, especially the power consumption (especially the power consumption). It is an effective means to improve the power efficiency of the receiver. The power consumption of the receiver is mainly composed of two parts: the Analog-to-digital Conversion, ADC and the baseband processing power, in which the power consumption of the ADC is exponentially related to the quantization precision, and is linear with the sampling frequency, and the baseband processing is used. The power consumption and the quantization precision are approximately square linear. With the development of the future communication system to the high bandwidth and high rate direction, the traditional high speed and high precision ADC will become an important constraint factor for the low power and low power system because of its huge power consumption and the consequent mass data processing. Therefore, the low precision quantization ADC is introduced to the number. In word receiver, it is considered as a means to effectively solve the complexity and power of large bandwidth and high rate receiver. It has received extensive attention from the academia and industry, and has important research significance and practical value.
In this paper, facing the urgent need of broadband and power efficiency optimization in future communication systems, the reception technology under low precision quantization is studied. The purpose is to provide a series of low-cost and low power solutions for large band communication. This paper is based on the design and performance analysis of the receiver, the influence of the system non rational factors, and the influence of the link code crosstalk three. The main research achievements and contributions are as follows:
In view of the design and performance analysis of the receiver, the design and performance analysis of the receiver in the low precision quantization communication system with orthogonal modulation are studied. First, the optimal single bit receiver under QPSK modulation is given, and the performance analysis is simplified with the distortion degree as the index. This paper further gives the suboptimal linear single ratio special connection. The simulation results show that the receiver has a signal-to-noise ratio of about 3dB (Signal-to-Noise-Rat) in the additive Gauss white noise channel (Additive White Gaussian Noise, AWGN). The simulation results show that the receiver has a signal to noise ratio (Signal-to-Noise-Rat) in the additive Gauss white noise channel (Additive White Gaussian Noise, AWGN) with a perfect channel state information (CSI). IO, SNR) loss, and only about 1dB SNR loss in dense multipath channels. This paper gives the optimal receiver under Z- quantization and obtains the bit log likelihood ratio. By Gauss approximation of the logarithmic likelihood ratio of bit, the performance analysis of the bit error rate (Bit Error Rate, BER) of the Z- order optimal receiver is given. The low precision is given. The relationship between the performance of the receiver and the quantization accuracy is quantified. The theoretical and simulation results show that when the quantized order is greater than 8, the performance gain obtained by increasing the quantization step is less than 0.1dB., and the effect of the method on the phase difference, the influence of the quantized threshold and the relationship between the performance gain and the quantization precision are highly consistent with the simulation results. The validity of the method is proved to provide design guidance for the practical application of a low precision quantization receiver. In addition, a practical linear piecewise receiver based on the minimum mean square error (Minimum Mean Square Error, MMSE) is given in order to reduce the complexity and increase the robustness of the receiver, and the corresponding letter is given. Channel estimation method.
In view of the influence of non ideal factors, this paper focuses on the effect of IQ imbalances on the QPSK monbant receiver and the corresponding countermeasures. It is proved that the amplitude imbalances have no effect on the single bimte receiver, while the phase unbalance and the common action of the transceiver phase difference will seriously deteriorate the performance of the single bit receiver. A single bimte receiver and a weighted double training training single bit receiver, which can effectively combat the performance loss caused by IQ imbalances in the AWGN channel and the sparse multipath channel, especially in the high SNR region. The strategy of the transmission end constellation rotation for the channel coding environment is proposed, and the phase diversity provided by the rotation strategy is provided. "It can provide a SNR gain of about O to 3dB for the convolutional codes with a rate of 1/2 to 7/8 in the AWGN channel and the sparse multipath channel. A 8- sector phase quantization receiver is proposed, and its robustness to the rotation phase is investigated. The implementation of a specific receiver in phase quantization is designed, so that the performance loss caused by IQ imbalances is completely eliminated. Because multipath can provide phase diversity, IQ imbalance has little impact on dense multipath channels.
In view of the influence of link Intercode crosstalk, the problem of link Intercode crosstalk in the process of applying low precision quantization receiver to high speed broadband system is studied. Using the channel characteristics of high speed broadband system, the Gauss approximation is carried out for Intercode crosstalk. The single specific connection under the Intercode Crosstalk Based on the Intercode crosstalk is given. The receiver is designed to perform the performance analysis of the receiver from the output signal to noise ratio loss and the BER two angles. The theoretical and simulation results show that the receiver has a performance loss less than 1dB when compared with no inter code crosstalk in the case of 250Mbps and below, indicating that no complex equalization technique is needed. A base for high intensity inter code crosstalk is proposed. The single bit time domain equalization algorithm for decision feedback and iterative demodulation can effectively combat the adverse effects of Intercode crosstalk. For the case of 500Mbps, the equalization algorithm has only about 1.2dB SNR loss relative to the case without Intercode crosstalk.
【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN851

【共引文献】