大MIMO中近似最大似然及格基约简检测算法研究

发布时间：2018-02-11 10:50

本文关键词： 大规模MIMO 低复杂度遗传LAS 基于LR的递归干扰消除基于LR的列表干扰消除　出处：《电子科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：大规模MIMO技术通过使用大量的接收天线服务多个终端和时分双工技术对现有通信系统有很大突破,大规模MIMO系统在丰富的散射环境下具有高频谱效率和高链路鲁棒的特点。实现大规模MIMO系统的重要困难在于检测问题,大规模MIMO系统中数据检测的计算代价太大,这种问题一直阻碍着大规模MIMO系统的实现。在过去二十年,很多检测方法被提出,目的均是为了减少计算复杂度,得到最佳检测性能。MIMO检测器主要分为两个类别。第一类检测器的复杂度取决于信道的特殊实现,例如树搜索,这些方法得到最佳性能的代价是复杂度呈指数倍,是复杂度最高的一类,如果使用一些终止条件得到的是次优的性能。另外一类检测器包含复杂度固定,不依赖信道实现的一类检测方法,这类从实现角度看更可取,因为它们消除大维度数据缓冲超出硬件能力的情况。本文的主要工作是针对这两类检测器分别研究两种不同的检测方法。本文研究的一种检测方法是遗传似然上升搜索(LAS)检测算法,该算法将遗传算法和LAS算法结合在一起,对LAS算法通过遗传的方式进一步处理,从而进一步提高大规模MIMO系统的接收性能,遗传LAS算法是LAS算法的一种改进方法,获得比LAS和禁忌搜索更好的检测性能。本文研究的另一种检测方法属于第二类检测技术,对于第二类检测技术,本文主要改进基于LR(Lattice Reduction)的递归干扰消除算法和基于LR的列表干扰消除算法两种方法。这两种算法都是根据矩阵的子空间能够比原矩阵的LR产生更加正交基向量的观点产生的。基于LR的递归干扰消除算法是一种部分干扰消除算法,它能够获得比基于LR的线性检测更好的检测性能,但是其复杂度比基于LR的线性检测稍高,而且它的复杂度还与递归次数有关。基于LR的列表干扰消除检测算法旨在将大规模MIMO系统分解为多个低维度MIMO系统,然后使用基于LR的线性检测检测每个低维度的MIMO系统,通过使用基于元素的LR算法,基于LR的列表干扰消除检测获得了复杂度和性能上的折中。
[Abstract]:Large scale MIMO technology has made a great breakthrough in the existing communication system by using a large number of receiving antennas to serve multiple terminals and time division duplex technology. Large-scale MIMO system has the characteristics of high spectral efficiency and high link robustness in rich scattering environment. The important difficulty of realizing large-scale MIMO system lies in the problem of detection. The computational cost of data detection in large-scale MIMO system is too high. This problem has been hampering the implementation of large-scale MIMO systems. In the past two decades, many detection methods have been proposed to reduce computational complexity. The complexity of the first kind of detector depends on the special implementation of the channel, such as tree search. The cost of these methods is exponential complexity. Is the most complex class, if some termination conditions are used to obtain sub-optimal performance, the other kind of detector includes a class of detection methods with fixed complexity and independent of channel implementation, which is preferable from the perspective of implementation. The main work of this paper is to study two different detection methods for these two kinds of detectors. One of the detection methods studied in this paper is genetic likelihood. Liter search lass detection algorithm, This algorithm combines genetic algorithm with LAS algorithm, and further deals with LAS algorithm through genetic method, thus further improving the receiving performance of large-scale MIMO system. Genetic LAS algorithm is an improved method of LAS algorithm. The detection performance is better than LAS and Tabu search. Another detection method studied in this paper belongs to the second kind of detection technology, and for the second kind of detection technology, In this paper, we mainly improve the recursive interference cancellation algorithm based on LR(Lattice reduction and the list interference cancellation algorithm based on LR. Both of these algorithms can produce more orthogonal basis vectors according to the subspace of the matrix than the LR of the original matrix. LR-based recursive interference cancellation algorithm is a partial interference cancellation algorithm. It can achieve better detection performance than LR-based linear detection, but its complexity is slightly higher than LR-based linear detection. LR-based list interference cancellation detection algorithm is designed to decompose large-scale MIMO systems into multiple low-dimensional MIMO systems, and then use LR-based linear detection to detect each low-dimensional MIMO system. By using element-based LR algorithm, LR-based list interference cancellation detection achieves a compromise in complexity and performance.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN919.3

【相似文献】