基于修正RZF的大规模MIMO预编码算法研究

发布时间：2018-05-30 09:51

本文选题：正则迫零预编码 + 大规模多输入多输出　；参考：《东南大学》2017年硕士论文

【摘要】：在大规模多输入多输出(MIMO,Multiple Input Multiple Output)系统中,随着用户数量的急剧增加,用户间干扰成为日益严重的问题。在下行链路中,对基站所发射的信号采用预编码技术可以有效地解决小区内的用户间干扰问题。预编码算法分为线性和非线性两种,正则迫零(RZF,Regularized Zero Forcing)预编码算法同时具备较低的计算复杂度和良好的编码性能,但是随着天线数增多,需要对高维矩阵进行求逆运算,从而导致它的复杂度急剧增加。因此,降低RZF预编码算法的复杂度成为研究热点。本学位论文主要研究具有低复杂度的修正RZF预编码算法。基本思路是通过对RZF预编码算法中的逆矩阵采用近似求解的方法来降低算法的复杂度。首先研究基于截短级数展开的RZF预编码技术。先后使用截短Neumann级数、截短Taylor级数和截短Kapteyn级数对逆矩阵进行展开和近似估计,从而得到相应的低复杂度RZF预编码算法。随后对基于截短Kapteyn级数展开的RZF预编码算法的系数进行优化,在保证复杂度不变的前提下,进一步提高了该类预编码算法的收敛速度。其次分析基于Newton迭代的RZF预编码技术,并用其对逆矩阵进行估计,然后分别从迭代初始值和迭代收敛速率两个方面对Newton迭代算法进行优化,进而得到基于Newton优化迭代的RZF预编码算法。最终,结合不同的应用场景给出了相应低复杂度的预编码方案。仿真结果表明:(1)在信道相干周期较大时,采用精确矩阵求逆方法的RZF算法能够同时获得最低的复杂度和最好的编码性能;(2)在信道相干周期较小时,迭代次数为2的基于Newton优化迭代的RZF预编码算法和截短阶数为2的基于Kapteyn级数展开系数优化后的RZF预编码算法,在保证了低于RZF预编码复杂度的同时,能够获得RZF预编码算法98%的性能;(3)在信道相干周期介于最小和最大范围内时,迭代次数为1的基于Newton优化迭代的RZF预编码算法和截短阶数不小于3的基于Neumann级数展开的RZF预编码算法能够获得RZF预编码算法70%的性能,但是前者的复杂度低于后者。
[Abstract]:In the large-scale MIMO (Multiple Input Multiple Output) system, with the rapid increase of the number of users, inter user interference becomes an increasingly serious problem. In the downlink, the pre coding technique for the signal transmitted by the base station can effectively solve the inter user interference in the cell. The precoding algorithm is divided into two parts. Linear and nonlinear two, the RZF (Regularized Zero Forcing) precoding algorithm has lower computational complexity and good coding performance. However, as the number of antennas increases, it is necessary to compute the inverse of the high dimensional matrix, which leads to the increase in the complexity of its complexity. Therefore, the complexity of the RZF precoding algorithm is reduced. This dissertation mainly studies the modified RZF precoding algorithm with low complexity. The basic idea is to reduce the complexity of the algorithm by using an approximate solution to the inverse matrix in the RZF precoding algorithm. First, the RZF precoding technique based on the truncation series expansion is studied. The truncated Neumann series, the truncated Ta is used successively. The ylor series and the truncated Kapteyn series carry out the expansion and approximate estimation of the inverse matrix, thus the corresponding low complexity RZF precoding algorithm is obtained. Then, the coefficients of the RZF precoding algorithm based on the truncated Kapteyn series expansion are optimized, and the convergence speed of the precoding algorithm is improved step by step on the premise of ensuring the complexity of the precoding. Secondly, the RZF precoding technology based on Newton iteration is analyzed, and the inverse matrix is estimated. Then, the Newton iterative algorithm is optimized from two aspects of the initial iteration value and the iterative convergence rate, and then the RZF precoding algorithm based on the Newton optimization iteration is obtained. Finally, the corresponding low complexity is given in combination with different application scenarios. The simulation results show that: (1) when the coherent period of the channel is large, the RZF algorithm using the exact matrix inversion method can obtain the lowest complexity and the best coding performance. (2) the RZF precoding algorithm based on the Newton optimization iteration and the truncated order of 2 at the time of the coherent period of the channel are 2. The RZF precoding algorithm based on the optimization of the Kapteyn series expansion coefficient can obtain the performance of the RZF precoding algorithm 98% while guaranteeing the complexity below the RZF precoding. (3) the RZF precoding algorithm based on Newton optimization iteration and the truncation number are not small when the channel coherent period is within the minimum and maximum range, and the number of iterations is 1. The RZF precoding algorithm based on Neumann series expansion in 3 can get the performance of RZF precoding algorithm 70%, but the complexity of the former is lower than that of the latter.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN919.3

【相似文献】