一种附加约束的ESPRIT测向方法

经典旋转不变技术（ESPRIT）算法测向估计性能在大多理论情况下接近于克拉美罗下界，且其在计算量方面要明显优于多信号分类算法（MUSIC）。本文首先论述了经典ESPRIT方法，描述了广义特征值分布在单位圆附近的实际工程应用情况，并提出了一种将广义特征值附加约束在单位圆上的修正ESPRIT算法。理论分析和仿真实验表明，该算法同经典的ESPRIT算法相比，当阵列存在几何误差的情况下，可明显改善测向估计的性能。

An Additional Constraint ESPRIT DOA Estimation Method

In most theoretical cases, the direction-finding performance of traditional Estimating Signal Parameter via rotational Invariance Techniques （ESPRIT） （ESPRIT） algorithm can be close to the CRLB （Cramer-Rao Low Bound）, and its computation complexity is much lower than the Multiple Signal Classification （MUSIC） algorithm. In this paper, we first discuss the traditional ESPRIT algorithm, and describe the practical generalized eigenvalues which are distributed in the vicinity of the unit circle. Then, we propose a modified ESPRIT algorithm that is based on an additional constraint for making the generalized eigenvalues on the unit circle. The theoretical analysis and simulation results demonstrate that the proposed method can improve the DOA estimation performance for the case of array geometric errors.