论文部分内容阅读
提出了特征空间法信源数估计方法,它将阵列信号的协方差估计值分别投影到信号的特征子空间和噪声的特征子空间。由于信号子空间与噪声子空间相互正交,易于由表征投影大小的判据值区分信号和噪声的贡献;本方法用的是M×M阶矩阵特征值分解,M为基元数,与波达方向估计用的相同,因此节省大量的计算量;它可以在实数空间中进行运算,进一步减少运算量。进行了数值计算,检验了判据值分布,以及在信源等功率、不等功率和空间相关色噪声等情况下特征空间法的性能。估计方法还用声纳数据进行了检验。所有这些结果均证明本估计方法性能优良。
A method of estimating the number of sources using eigenspace method is proposed. It projects the covariance estimates of the array signals respectively into the signal feature subspace and the noise feature subspace. Because signal subspace and noise subspace are orthogonal to each other, it is easy to distinguish the contribution of signal and noise by the criterion value that characterizes the projection size. This method uses M × M matrix eigenvalue decomposition, M is the number of primitives, The same direction estimation is used, thus saving a lot of computation; it can be operated in real space to further reduce the computational complexity. The numerical calculation is carried out to verify the distribution of the criterion value and the performance of the eigenspace method under the conditions of power, unequal power and spatially correlated color noise. The estimation method was also tested with sonar data. All of these results prove that this method is superior in performance.