【作者】 王春梅；

【导师】 吕锋；

【作者基本信息】 武汉理工大学 ， 通信与信息系统， 2013， 硕士

【摘要】 目前LTE (Long Term Evolution,长期演进)被全球移动通信产业寄予厚望,由于LTE系统在发送端采用了MIMO技术,在接收端,每个天线接收到的是多个频率和时间上都重叠的发射信号的叠加,因此接收端必须采用信号检测算法才能分辨并恢复出各个发射天线上的信号。信号检测算法的性能好坏决定了接收端能否判决出正确的发送信号,对整个物理链路的性能有至关重要的影响。目前,针对MIMO技术的信号检测算法已经有很多,各检测算法的性能与复杂度各有不同。然而,传统的信号检测算法的误码率性能或运算复杂度并不十分理想,难以找到一种既具有良好的性能又不需要较高的运算复杂度的检测算法。因此,对于LTE系统的信号检测技术的研究是十分有实用意义的。本文的主要工作如下。(1)给出LTE下行信道的处理流程和LTE系统中基于空间复用的LTE系统模型,并依此设计与搭建LTE系统仿真平台。(2)对几种经典的信号检测算法进行了复杂度与仿真性能分析,包括最大似然算法、迫零算法、最小均方误差算法、基于干扰删除的VBLAST算法、QRD-M算法和SD算法。(3)提出了一种基于信噪比的分组检测算法,分析了该改进算法的运算复杂度并在仿真平台上进行了性能仿真验证。该改进算法是SD检测算法的基础上,引入了干扰抵消和最小欧式距离的概念,该算法的思想是,按接收信噪比的大小重新排列信道响应矩阵,设定一个分组值K,最先检测的K层信噪比较大,采用SD算法进行检测,随后检测的Nt-K层信噪比较低,在消除完已判决的K个符号在接收信号向量中的影响后,接下来的每层进行干扰抵消判决,选取最大似然解。性能仿真证明通过设定合理的K值,改进算法确实能取得较好的性能,并且能在复杂度与性能中取得较好的折中。(4)提出了一种基于QRD_M算法软输出算法,在仿真平台上验证了在添加Turbo编码后,软输出算法的性能优于硬判决检测算法。软输出QRD-M检测算法首先改进了QRD-M算法,使其复杂度有了一定程度上的降低,然后将改进算法与软信息的输出结合起来,使得在信号检测的过程中减少信息损失,从而提高了整体的检测性能。更多还原

【Abstract】 At present, LTE is inspired high hopes by global mobile communications industry, the LTE system adopts the MIMO technology at the transmitting end, at the receiving end, each antenna received the superposition of emission signal which are overlapping in frequency and time, so the receiving end must adopt signal detection algorithm to identify and recover the transmitted signal. The detection algorithm’s performance determines whether the receiving end can get the correct transmission signal, so signal detection technologies have a critical influence on the performance of the entire physical link in the LTE system. Currently, there have been many signal detection algorithms for the MIMO system, and they are different in performance and complexity. However, the conventional signal detection algorithms are not available for LTE system because of the constraints of the BER performance and computational complexity. Therefore, the research on the signal detection technology for the LTE system is very practical and significant. The major work of thesis is as follows.(1) This thesis shows the spatial multiplexing model of LTE system and the processing flow of the LTE downlink channel, which laid a good theoretical foundation for the design and construction of the practical LTE simulation platform.(2) According to simulation results, this thesis compares the performance and complexity of several classic signal detection algorithms based on spatial multiplexing, including maximum likelihood algorithm, zero forcing algorithm, the minimum mean square error algorithm, the VBLAST algorithm, QRD-M algorithm and SD algorithm.(3) This thesis designs a packet detection algorithm based on signal-to-noise ratio, then analyzes its computing complexity and implements the improved algorithm in the LTE simulation platform. This algorithm is based on sphere decoding algorithm and introduces the concept of interference cancellation and minimum Euclidean distance. The modified algorithm firstly rearranges the channel response matrix according to Post-Detection SNR, then set the first K detection layers have larger Post-Detection SNR and higher reliability, the subsequent NT-K detection layers have lower Post-Detection SNR and poor reliability. For the first detected K symbols, it adopts the SD algorithm, for the next Nt-K layers, firstly it eliminates judgment K symbols in the received signal vector, then the each layer judges M estimated values with minimum Euclidean distance after interference cancellation, finally selectes the maximum likelihood value as transmited vector. According to simulation results, it proves that the improved algorithm can obtain good performance and achieve the expected tradeoff between performance and complexity.(4) This thesis puts forward an improved soft-QRD_M algorithm and analyzes its computational complexity and implements it on simulation platform. This algorithm makes an improvement about QRD-M algorithm to reduce its complexity and then extract the soft information which can reduce information loss in the decoding process. According to the simulation results, in turbo coding platform, this algorithm had better performance than hard detection algorithm.更多还原