一种新的DOA估计的高分辨率算法

New high-resolution DOA estimation algorithm

波达方向估计(DOA)在阵列信号处理中非常重要,而传统算法如多重信号分类算法(MUSIC)需要对阵列接收数据的协方差矩阵进行特征分解,并在全空域进行谱峰搜索,运算量巨大,尤其是二维DOA估计算法还存在稳健性较差的问题。提出了一种基于最小互熵谱分析的DOA算法,该算法只需要一次较少阵元的阵列采样(快拍)数据即可得到具有高分辨率的谱分析结果。进一步采用倒谱法,通过逆FFT变换提高了最小交叉熵谱分析算法的收敛速度。数值仿真结果表明,该算法较常规空间谱估计方法有更高的分辨力和更小的运算量,能够对阵列信号进行实时处理。算法不依赖于预先估计的信源数目,具有较好的宽容性,较高的分辨力以及极低的旁瓣电平。

DOA estimation is very important in array signal processing, while traditional algorithms such as Multiple Signal Classification algorithm（MUSIC） requires eigenvalue decomposition of covariance matrix and spatial spectral peak in the whole search, which is difficult and time-consuming. In particular, there exists robustness problem in two-dimensional DOA estimation algorithm. This paper presents a DOA estimation algorithm based on minimum cross-entropy spectral analysis, which requires only one array of sampling（snapshot）data with fewer elements. And it still can obtain high-resolution spectral analysis. Based on cepstrum method and inverse FFT transform it can improve the convergence speed. Numerical simulation results show that compared to the conventional spatial spectrum estimation methods this algorithm has a higher resolution and a smaller amount of computation, being capable of real-time array signal processing. The algorithm does not depend on the number of pre-estimate of the source, results in good tolerance, high resolution and low side lobe level.