摘要
将马尔可夫蒙特卡罗(MCMC)方法与多重子空间分类(MUSIC)方法估计相结合,提出一种用于联合估计多个目标的频率、方位和俯仰,基于吉布斯抽样的MUSIC多维参数联合估计新方法。该方法将MUSIC方法的谱函数作为频率、方位和俯仰的联合概率密度函数,采用MCMC吉布斯抽样方法对该联合概率密度函数进行采样。理论分析和仿真实验表明:在目标个数较少时,该方法不仅保持了常规MUSIC方法的高分辨能力。
A new Markov Chain Monte Carlo（MUSIC） multi-parameters joint estimator based on Gibbs sampling （Gibbs-MUSIC） is proposed to jointly estimate the frequency,directions and elevations of multiple sources. The method regards the power of MUSIC spectrum function as target distribution up to a constant of proportionality, and uses Gibbs sample, one of the most popular MCMC technique, to sample from it. Theoretical analysis and simulations demonstrate that the new method not only possesses the performance of high-resolution in conventional MUSIC method but also provides a less computation and storage costs than that of conventional MUSIC method under the condition where the number of signal sources is small.
出处
《传感器与微系统》
CSCD
北大核心
2008年第6期62-65,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(60572098)
国家"863"计划资助项目(2007AA01Z478)
河南省自然科学基金资助项目(0511014300)
