Abstract: In order to reduce the computational complexity and resolution threshold of classical TCT algorithm for wideband signals，the real value transforming process and an improved MUSIC method based on subspace projection are introduced for central symmetric arrays to propose a new direction finding method for wideband sources.In this method，firstly，noise removal and forward-backward smoothing are conducted on the data covariance matrices at different frequencies，then these matrices are transformed into real ones by left-real transforming matrix，which achieves a substantial reduction of computational load in later eigen-decomposition operations，next focusing matrices for different frequencies are computed by conventional TCT method to transform data covariance matrices into ones at a common reference frequency，finally the direction of sources are obtained by a series of one-dimensional search of the maxima of an improved MUSIC estimator based on subspace projection.Simulations show that the proposed method reduces not only the computational overhead but also the resolution threshold and bias noticeably compared with conventional TCT method.

Key words: wideband sources, real value processing, focusing matrices, subspace projection, computational complexity

摘要： 为了降低宽带信号TCT方位估计算法的运算量和分辩门限，针对中心对称阵列，将实值处理过程和子空间投影MUSIC的思想引入宽带信号方位估计，提出一种宽带信号方位估计新方法。该方法首先各个频率子带的数据协方差矩阵去噪且双向平滑，然后使用左实变换矩阵将它们变换为实数矩阵，从而大大降低了后面矩阵特征分解时运算量，接着使用双边相关变换（TCT）方法选取聚焦矩阵对各个频率子带的实协方差矩阵进行变换得到同一参考频率点的数据协方差阵，最后利用基于子空间投影的MUSIC方法进行一系列一维搜索算法来求得各个目标的方位角。仿真实验结果表明，此方法比常规宽带TCT方法运算量小，且具有更低的分辨门限和更小的估计偏差。

关键词: 宽带信号, 实值处理, 聚焦矩阵, 子空间投影, 计算复杂度