【摘要】 常规多信号分类(MUSIC)在估计信号或噪声子空间时未利用阵列的方向矢量信息。为改善波达方向(DOA)估计性能,提出一种新的角信号子空间概念。首先,由Gram行列式和超维空间中多面体体积公式,给出常规MUSIC方法的几何解释。其次,利用阵列响应矢量扩展观测数据矩阵,在每个搜索方向由增广数据矩阵的奇异值分解获得角信号子空间估计。理论分析表明,常规MUSIC零谱相当于超维空间中由阵列观测数据矢量和搜索方向矢量决定的多面体体积。仿真实验表明,利用角信号子空间能够较明显地改善DOA估计性能,特别是信号相关、信噪比较低以及快摄数较小的情况。更多还原

【Abstract】 Information on the steering ve ctor is not exploited when estimating signal or noise subspaces in the standard multiple signal classification (MUSIC) method. The direction of arrival (DOA) estimate was improved by using the propo sed angular subspace concept. The analysis is based on the geometrical interpret ation of the MUSIC method derived from the formula for the Gram determinant and the volume of the polygon. The angular signal subspace was then calculated from the singular value decomposition of a steering-vector-augmented data matrix in each direction. Theoretical analysis showed that the null spectrum of the ordin ary MUSIC method is equal to the volume of the polygon, determined from the data matrix and a steering vector. Simulation results show that the DOA estimate per formance can be improved by using the angular signal subspace, especially with c orrelated signals, relatively low signal-to-noise ratios and a small number of snapshots.更多还原