针对多输入多输出(Multi-Input Multi-Output,MIMO)阵列多目标定位,提出一种基于子空间特征分解的MIMO阵列旋转不变子空间算法(MIMO array estimation of signal parameters via rotational invariance techniques,简称MIMO-ESPRIT)。MIMO阵列各阵元发射彼此独立信号,因此阵列输出数据协方差矩阵不存在降秩以及信号子空间向噪声子空间扩散的现象,可直接应用子空间分解算法进行MIMO阵列目标方位估计。性能分析和仿真结果表明,随着发射阵元个数的增加和发射阵元间距的扩大,算法的多目标分辨能力和方位估计精度将得到明显改善。
Based on subspace eigen decomposition,MIMO array estimation of signal parameters via rotational invariance techniques is proposed for multiple targets localization using MIMO array.Because MIMO array sensors transmit mutually orthogonal signals,a rank deficiency is nonexistent in the array output covariance matrix and a divergence of a signal eigenvector into the noise subspace will not happen,which contributes to direct application of subspace eigen decomposition to multiple targets direction of arrival(DOA) estimation of the MIMO array.The simulation results show that increasing the number of transmit sensors and extending their inter-element space will result in superior multiple targets DOA estimation performance of MIMO-ESPRIT.