基于LLR的低复杂度MASSIVE MIMO系统信号检测算法
发布时间：2018-12-13 06:11
【摘要】：Massive MIMO技术扩大系统天线配置达几十甚至上百,天线数的量变使系统获得了特性方面的质变,在很大程度上提升了无线传输的频率利用率、信道容量及可靠性。与此同时,天线数量较多,传统的接收端的信号处理方式检测性能较差,而且过于复杂不利于硬件实现。所以主要研究适用于Massive MIMO系统的低实现复杂度的信号检测算法。首先对比了点到点MIMO以及多用户MIMO的系统模型,在此基础上扩大天线配置,引出了Massive MIMO系统模型,并说明了Massive MIMO技术的实际优势以及实际实现中的瓶颈。信号检测方面,整理了常用的线性检测与非线性检测方法,指出其适应的通信场景以及不足。鉴于基站天线数远大于发射端天线数时,传统的线性检测即可获得较好的性能,所以课题主要研究收发两端天线配置较均衡时的信号检测算法。为了进一步提高检测精度,研究算法都是以MMSE检测算法为起始点,不同于传统的硬判决,几种算法目标信号的确定都是根据信息比特的对数似然比。其次在一维显著特征向量搜索检测算法的基础上提出了改进的多维显著特征向量搜索检测算法。一维特征向量搜索算法假设信号最可能存在在某一显著特征向量方向上,针对选定的特征向量进行遍历从中选出最优方向的信号。在此基础上,多维特征向量搜索综合多个特征向量的结果进行信号的判决不仅可获得更优的检测性能而且可以大大减少候选信号的个数。根据复杂度统计结果,多维搜索的复杂度始终低于一维搜索。所以与其一维搜索算法相比,改进的多维特征向量搜索检测获得了较好的检测性能以及较低的硬件实现复杂度。最后研究了基于检测误差建模的天线选择检测算法。该算法设某天线集合上的信号恰为某星座点集合,最终求得使检测误差最小的目标信号向量。若选取天线的个数为1,则为单天线选择检测算法,若为多个则为多天线选择检测算法。低信噪比时,单天线选择检测的检测性能即优于本文其他算法,高信噪比时,多天线选择检测算法在复杂度提高较少时仍能维持较好的检测性能。仿真对比可发现单天线选择检测算法的复杂度能较好的适应天线配置的改变,实现低复杂度的Massive MIMO信号检测算法。
[Abstract]:The Massive MIMO technology expands the antenna configuration of the system to tens or even hundreds. The quantitative change of the antenna number makes the system obtain the qualitative change of characteristic, which greatly improves the frequency utilization ratio, channel capacity and reliability of wireless transmission. At the same time, the number of antennas is large, the detection performance of the traditional signal processing method is poor, and the complexity is not conducive to hardware implementation. Therefore, a low complexity signal detection algorithm suitable for Massive MIMO system is studied. Firstly, the system models of point-to-point MIMO and multi-user MIMO are compared. On this basis, the antenna configuration is expanded, and the Massive MIMO system model is introduced, and the practical advantages of Massive MIMO technology and the bottleneck in practical implementation are explained. In the aspect of signal detection, the common linear and nonlinear detection methods are sorted out, and the suitable communication scenes and their shortcomings are pointed out. Because the base station antenna number is far larger than the transmitter antenna number, the traditional linear detection can obtain better performance, so the thesis mainly studies the signal detection algorithm when the antenna configuration is more balanced. In order to further improve the detection accuracy, the algorithms are based on MMSE detection algorithm as the starting point, which is different from the traditional hard decision. The determination of the target signal of several algorithms is based on the logarithmic likelihood ratio of information bits. Secondly, an improved multi-dimensional salient feature vector search detection algorithm is proposed on the basis of one-dimensional salient feature vector search detection algorithm. The one-dimensional eigenvector search algorithm assumes that the signal is most likely to exist in the direction of a salient eigenvector and traverses the selected eigenvector to select the optimal direction of the signal. On this basis, multi-dimensional eigenvector search synthesizes the result of multi-eigenvector to judge the signal not only can obtain better detection performance, but also can greatly reduce the number of candidate signals. According to the statistical results of complexity, the complexity of multidimensional search is always lower than that of one-dimensional search. Compared with the one-dimensional search algorithm, the improved multi-dimensional eigenvector search algorithm has better detection performance and lower hardware implementation complexity. Finally, the antenna selection detection algorithm based on error modeling is studied. In this algorithm, the signal on a set of antennas is exactly a set of constellation points, and finally the target signal vector with minimum detection error is obtained. If the number of antennas selected is 1, it is a single antenna selection detection algorithm, and if more than one antenna is selected, it is a multi-antenna selection detection algorithm. When the SNR is low, the detection performance of single antenna selection detection is better than that of other algorithms in this paper. When the SNR is high, the multi-antenna selection detection algorithm can maintain better detection performance even when the complexity is less improved. The simulation results show that the complexity of the single antenna selection detection algorithm can adapt to the antenna configuration change and realize the low complexity Massive MIMO signal detection algorithm.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.3
本文编号：2376041
[Abstract]:The Massive MIMO technology expands the antenna configuration of the system to tens or even hundreds. The quantitative change of the antenna number makes the system obtain the qualitative change of characteristic, which greatly improves the frequency utilization ratio, channel capacity and reliability of wireless transmission. At the same time, the number of antennas is large, the detection performance of the traditional signal processing method is poor, and the complexity is not conducive to hardware implementation. Therefore, a low complexity signal detection algorithm suitable for Massive MIMO system is studied. Firstly, the system models of point-to-point MIMO and multi-user MIMO are compared. On this basis, the antenna configuration is expanded, and the Massive MIMO system model is introduced, and the practical advantages of Massive MIMO technology and the bottleneck in practical implementation are explained. In the aspect of signal detection, the common linear and nonlinear detection methods are sorted out, and the suitable communication scenes and their shortcomings are pointed out. Because the base station antenna number is far larger than the transmitter antenna number, the traditional linear detection can obtain better performance, so the thesis mainly studies the signal detection algorithm when the antenna configuration is more balanced. In order to further improve the detection accuracy, the algorithms are based on MMSE detection algorithm as the starting point, which is different from the traditional hard decision. The determination of the target signal of several algorithms is based on the logarithmic likelihood ratio of information bits. Secondly, an improved multi-dimensional salient feature vector search detection algorithm is proposed on the basis of one-dimensional salient feature vector search detection algorithm. The one-dimensional eigenvector search algorithm assumes that the signal is most likely to exist in the direction of a salient eigenvector and traverses the selected eigenvector to select the optimal direction of the signal. On this basis, multi-dimensional eigenvector search synthesizes the result of multi-eigenvector to judge the signal not only can obtain better detection performance, but also can greatly reduce the number of candidate signals. According to the statistical results of complexity, the complexity of multidimensional search is always lower than that of one-dimensional search. Compared with the one-dimensional search algorithm, the improved multi-dimensional eigenvector search algorithm has better detection performance and lower hardware implementation complexity. Finally, the antenna selection detection algorithm based on error modeling is studied. In this algorithm, the signal on a set of antennas is exactly a set of constellation points, and finally the target signal vector with minimum detection error is obtained. If the number of antennas selected is 1, it is a single antenna selection detection algorithm, and if more than one antenna is selected, it is a multi-antenna selection detection algorithm. When the SNR is low, the detection performance of single antenna selection detection is better than that of other algorithms in this paper. When the SNR is high, the multi-antenna selection detection algorithm can maintain better detection performance even when the complexity is less improved. The simulation results show that the complexity of the single antenna selection detection algorithm can adapt to the antenna configuration change and realize the low complexity Massive MIMO signal detection algorithm.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.3
【参考文献】
相关期刊论文 前2条
1 杨大江;钟子发;宋常建;葛梅宝;;一种用于大规模MIMO信号检测的改进算法[J];重庆邮电大学学报(自然科学版);2015年02期
2 张源;沈皓;赵昆;夏斌;;LTE-A上行链路中单用户MIMO的球形译码算法性能分析[J];电讯技术;2015年01期
相关硕士学位论文 前2条
1 杨海洋;Massive MIMO中低复杂度接收算法的研究[D];电子科技大学;2013年
2 史开新;MIMO系统中信号检测技术的研究[D];西安电子科技大学;2008年
，本文编号：2376041
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2376041.html
