论文部分内容阅读
在相干分布式非圆(CDNC)信号波达方向(DOA)估计中,针对阵列输出矩阵扩展后维数增加带来的较大运算量问题,基于降维的多级维纳滤波(MSWF)技术,引入回溯优化思想,提出了一种快速估计算法。该算法首先利用信号非圆特性扩展阵列输出矩阵,然后通过MSWF递推分解快速求出信号子空间,避免了计算阵列协方差矩阵及特征分解,并且在递推过程中引入回溯优化机制提高了各级匹配滤波器的估计性能,最后由最小二乘(LS)或者总体最小二乘(TLS)得到DOA估计。仿真分析表明,所提算法与相干分布式非圆信号旋转不变子空间算法(CDNC-ESPRIT)性能相当,但复杂度得到了大幅度降低,相比于基于MSWF的非圆信号快速子空间(NC-MSWF-FS)算法,在较小的复杂度代价下大幅度提升了低信噪比时的估计性能,并且对初始参考信号的选取具有了较强的鲁棒性。
In the DOA (DOA estimation) of coherent distributed non-circular (CDNC) signals, aiming at the large computational complexity caused by the increase of array output matrix dimension, the multi-level Wiener Filter (MSWF) , Introduce backtracking optimization idea, put forward a fast estimation algorithm. Firstly, the output matrix of the array is extended by using the non-circular nature of the signal, and then the signal subspace is obtained by recursive decomposition of the MSWF. The array covariance matrix and eigendecomposition are avoided, and the backtracking optimization mechanism is introduced to improve the performance of each Level matched filter, and finally the DOA estimation is obtained by Least Squares (LS) or Total Least Squares (TLS). Simulation results show that the proposed algorithm has the same performance as CDNC-ESPRIT, but the complexity is greatly reduced. Compared with the non-circular signal fast subspace based on MSWF NC-MSWF-FS algorithm, which greatly improves the estimation performance at low signal-to-noise ratio with less complexity and robustness to the initial reference signal selection.