由频率分集阵列(FDA)和多输入多输出(MIMO)技术组成的FDA-MIMO雷达因其可以形成角度-距离相关的波束而广受关注。相比于经典的子空间谱估计算法,Capon算法无需特征值分解,计算量小,不用预估信号个数,估计性能稳定。文中提出一种应用于FDA-MIMO雷达的实值降维Capon算法。该算法使用前后项平滑技术和酉变化把接收信号变为实值矩阵,再利用矩阵直积的性质将实值2D-Capon降至一维,用两个一维搜索分别估计角度和距离,极大降低计算量。经过仿真验证,该算法对相干与非相干目标都能准确估计,且保持与二维多重信号分类(2D-MUSIC)算法几乎相同估计精度的同时,有着较低的计算复杂度。

Combining frequency diverse array(FDA)and multiple-input multiple-output (MIMO) technology to form the so-called FDA-MIMO radar has received wide attention because of its range-angle-dependent beam-pattern. Compared with the classic spatial spectrum estimation algorithms, Capon method has lower computational complexity for deserting eigenvalue decomposition and more stable estimation performance without determining the number of targets. A real-valued and reduced-dimension Capon algorithm applied to FDA-MIMO is proposed. Firstly, an extended method with the symmetric structure and unitary transformation is adopted to transform receiving matrix from complex domain to real domain. After that, transform the real-valued 2D-Capon to one-dimension by utilizing direct product of matrix, estimating angle and range through multiple one-dimension respectively which extremely reduce computational complexity. According to computer simulation results, the proposed algorithm, having an accurate estimation of coherent and incoherent targets, not only retains almost the same estimation performance as two dimension multiple signal classification (2D-MUSIC) algorithm, but also preserves lower computational complexity.

TN957