基于Gibbs采样的贝叶斯多目标高分辨方位估计  

Baysian DOA Estimator by Gibbs Sampling

作　　者：孙毅[1] 刘科伟[1] 黄建国[1] 褚福照[1] 

机构地区：[1]西北工业大学航海工程学院,西安710072

出　　处：《计算机工程与应用》2002年第12期36-38,共3页Computer Engineering and Applications

基　　金：航空基金资助项目(编号:97F53062)

摘　　要：讨论了一种利用Gibbs采样方法的贝叶斯多目标高分辨定向新方法(BDGS),给出了这种贝叶斯定向新方法的推导过程。该方法将一种新方法Gibbs采样应用于贝叶斯高分辨方法的计算过程。新方法不仅具有贝叶斯多目标高分辨定向方法的优越性能,而且还降低了原贝叶斯方法计算的复杂程度。并在最后给出了新方法的性能分析,与MUSIC(子空间法)和MLE(最大似然法)的比较结果表明新方法有更高的分辨率,在低信噪比情况下性能也较好。A new Bayesian DOA estimator by Gibbs sampling(BDGS)is proposed to estimate the directions of multiple sources.The estimator uses the Gibbs sampler in the computation of the Bayesian high-resolution method.The formulation of the proposed Bayesian DOA estimator is derived.The new method not only possesses the performance of high -resolution direction finding but also provides reduced computational complexity to the original Bayesian method.Comparison with MUSIC and maximum likelihood estimator(MLE)shows that the new estimator has higher resolution.In addition,it is more robust for low SNRs.

关 键 词：GIBBS采样 多目标高分辨方位估计 贝叶斯方法 

分 类 号：TN911.7[电子电信—通信与信息系统]