一种多维MUSIC方位估计的快速实现方法

A Fast Method of Multidimensional MUSIC DOA Estimation

多维MUSIC方法在估计波达方向方面有着优良的性能,但由于计算量大而难以实时应用。提出了多维MUSIC快速方法(ACOMD-MUSIC)。首先将搜索空间按照均匀分布进行抽样,选择一组具有加权值的高斯核函数作为信息素概率密度函数对搜索空间进行抽样,直至迭代结束得到最优解,将信号子空间与阵列流型的拟合关系作为目标函数得到ACOMD-MUSIC方法。仿真结果表明,ACOMD-MUSIC方法在保持了原方法高分辨性能的同时将计算量由O(((θ_h-θ_l)/a)~p)减少到O(P×I+L)。

Abstract：For multidimensional MUSIC algorithm for computationally intensive and for engineering problems difficulty, this paper propose fast estimation algorithm. In this method, a Gaussian kernel as is defined to accomplish the sampling process, and the maximum of objective function can be obtained after reasonable iterations. ACOMD-MUSIC algorithm is got by the signal subspace and array manifold vector fitting relationship as the objective function. Simulation results show that the three fast al_ gorithms maintain excellent performance of the original algorithm, while the computational complexity is significantly reducing smaller than that of the original algorithm, especially when the number of sources increases. ACOMD-MUSIC algorithm decre- ases the calculation of MD-MUSIC from O（（（θ_h-θ_l）/a）~p）toO（P×I＋L）.