训练样本不足时的子空间信号检测方法

摘要

In order to overcome the difficulty of detecting a subspace signal with insufficient training data, two effective reduced-rank subspace detectors are proposed. According to the theory of principal compo-nent analysis(PCA),the sample covariance matrix(SCM),contained in conventional detection statistic,is replaced by the production of the noise eign-subspace and its conjugate transpose. This results in reduced-rank subspace detectors. To further improve the robustness,the matrix inversion operation is substituted by the Moore-Penrose inversion. The comparison with conventional detectors shows that the proposed re-duced-rank subspace detectors can provide improved detection performance,no matter the number of the training data is sufficient or not.%为了解决训练样本不足时的子空间信号检测问题,提出了两种有效的降秩检测器.基于主分量分析(PCA)的思想,先把常规自适应子空间检测器中采样协方差矩阵(SCM)的求逆运算用噪声特征子空间矩阵与其共轭转置的乘积代替,构造降秩子空间检测器;为进一步提高算法稳健性,把降秩子空间检测器的求逆运算用Moore-Penrose逆代替.仿真结果表明,所提方法在训练样本充足及不足时,均比现有方法具有更好的检测性能.