基于相干对消的噪声子空间自适应估计

MUSIC算法的运算量主要集中在特征值分解和空间谱的搜索两部分，为避免进行特征 值分解，将相干信号自适应对消的思想用于噪声子空间的估计。由于阵元输出信号完全相干 ，视一个阵元输出为参考信号，其它阵元输出能够完全对其进行对消，得到的系数矩阵作为 噪声子空间的估计。基于LMS算法，给出了算法的矩阵形式，得到的噪声子空间估计算法的 运算量大大降低，且以迭代的方式进行，适合应用于运动信号源的跟踪，在阵元数较大时能 很好地逼近MUSIC算法性能。为减少空间谱搜索过程的计算量，对搜索过程先用系数矩阵的 一列进行搜索，然后采用其它列对搜索峰值进行验证。仿真结果显示算法具有很好的空间谱 估计性能和DOA跟踪性能。

Adaptive Noise Subspace Estimation Based on Coherent Signal Canceller

Most of the computational complexities of MUSIC algorithms are centered in eigendecomposition and full dimensional DOA search. To avoid eigendecomposit ion, adaptive coherent signal cancellation algorithm is introduced to estimate noise subspace. Because all outputs of the arrays are coherent, each one can be cancelled by others, and this process results in a coefficient matrix to be the estimation of noise subspace. The matrix form of this method is provided based on LMS algorithm, worki ng in parallel and iteratively which reduces the computational complexity with l ittle performance loss when the array number is not too small contracted to MUSI C algorithm and suits engineering implementation and DOA tracking. The spectrum peaks are first searched by one vector of the matrix, and the other vectors are used t o identify all the peaks. Using this method, the DOA search process is optimized too. The simulation analysis indicates that this novel algorithm is active.