针对传统算法在进行波达方向角估计时需要进行全空间的谱峰搜索,且依赖初始化迭代和插值的缺陷,设计了一种贝叶斯估计方法。首先,构建基于均匀圆阵的频域信号处理模型,提出了基于马尔科夫链蒙特卡罗进行波达方向角估计的方法,实现了无需初始化和插值的测向算法,且不需要进行全空间的谱峰搜索即可直接估计结果,同时推导了该模型下进行波达方向角估计的克拉美罗下界。在以测向结果为先验信息,同时设计了一种新的空域滤波增益模型。利用马尔科夫链蒙特卡罗算法对信号子空间进行解耦合,然后计算滤波器增益系数,以此实现空域滤波。最后在无线电设备搭建的真实环境中进行实验,验证了基于马尔科夫链蒙特卡罗算法进行阵列波达方向角估计和空域滤波的有效性,并且与其他模型相比要明显优于后者。

Aiming at the shortcomings of traditional algorithms that need to perform full-space spectral peak search when estimating direction of arrival, and relying on initialization iteration and interpolation, a Bayesian estimation method is designed in this paper. Firstly, a frequency-domain signal processing model based on uniform circular array is constructed, and a method for direction of arrival estimation based on Markov Chain Monte Carlo is proposed, which realizes a direction-finding algorithm that does not require initialization and interpolation, and the results can be directly estimated without the need to perform a full-space spectral peak earch. At the same time, the Cramer-Rao lower bound for direction of arrival estimation under this model is deduced. Based on he direction-finding results as prior information, a new spatial filtering gain model is designed. The signal subspace is decoupled by the Markov Chain Monte Carlo algorithm, and then the filter gain coefficient is calculated to realize the spatial filtering. Finally, experiments are carried out in the real environment built by the radio equipment, and the effectiveness of the array direction of arrival estimation and spatial filtering based on the Markov Chain Monte Carlo algorithm is verified, and compared with other models, it is significantly better than the latter.

TN957. 52

国家重点研发计划资助项目(2020YFB1711902);四川省科技计划资助项目(2020SYSY0016)