基于空间虚拟延迟抽头的信源数检测方法

Source Number Detection Methods Based on the Spatial Virtual Tapped-delay-line

基于空间虚拟延迟抽头技术,给出一种空时协方差矩阵,论述了阵列协方差矩阵的特征向量与空时协方差矩阵的正交原理,引入一种特征向量空时投影的概念。给出了两种利用这种特征向量空时投影的信源数检测方法。一种是基于启发式算法的,与GDE方法类似,称为类GDE方法;另一种是基于Bootstrap算法的,称为BBE方法。理论分析显示,两种新的检测方法,不再依赖于阵列协方差矩阵的特征值,无需对信号和噪声的分布作出假设,具有很好的抗空间非均匀噪声能力。同时,类GDE在低信噪比条件下性能更好,而BBE则无需人为设置检测门限,具有较好的工程实用价值。基于仿真数据和湖试数据的测试结果证明了所提方法的有效性。

Based on the spatial virtual tapped-delay-line technique, the concept of spatiotemporal co- variance matrix （STCM） was introduced. At the same time, the concept of projection of eigenvectors of the array covariance matrix onto column-vectors of the STCM was also discussed. The two ap- proaches, based on Bootstrap and heuristic algorithms, were proposed for developing source number detection methods. The Like Gerschgoin disk estimator （GDE） method was proposed by the projec- tions of eigenvectors and the heuristic algorithm. It avoids the drawback of the existing source number detection methods based on eigenvalues, can work in the case of lower signal-to-noise ratio（SNR） and unknown spatially nonuniform noise. As no assumption is made on the distribution of the noise or the signals, another source number detection method was proposed by the projections of eigenvectors and the Bootstrap algorithm. In the latter method, no subjective detection threshold is required. The ex- periments with data from computer simulations as well as lake trials show that proposed methods are effective and superior to existing methods in different scenarios.