English

应用科学学报

• 论文 • 上一篇    下一篇

非高斯噪声参数估计的马氏链蒙特卡罗法

张曙霞 ，蒋宇中 , 徐大勇   

  1. 海军工程大学 电子工程学院 ，湖北 武汉430033
  • 收稿日期:2007-03-30 修回日期:2007-06-12 出版日期:2007-11-30 发布日期:2007-11-30

Estimation of Non-Gaussian Noise Parameters Using Markov Chain Monte Carlo Method

ZHANG Shu-Xia, JIANG Yu-Zhong , XU Da-yong   

  1. College of Information and Electrical Engineering, Naval University of Engineering, Wuhan 430033, China
  • Received:2007-03-30 Revised:2007-06-12 Online:2007-11-30 Published:2007-11-30

被引次数

4

摘要： 基于马氏链蒙特卡罗法(MCMC), 提出一种快速收敛特性的A类噪声模型参数贝叶斯估计算法。区别于传统的参数估计算法，它不仅可以估计脉冲指数A、高斯脉冲功率比 、噪声平均功率 ，还可以估计信道隐含状态。尽管该算法运算量大，但优点是所需样本少、能实现整体优化及并行处理。计算机仿真验证了该方法的有效性。

关键词: 非高斯噪声, A类噪声, 脉冲噪声

Abstract: A fast convergence Bayesian estimator of the class A model parameters is derived and calculated using the Markov Chain Monte Carlo (MCMC) procedure. This estimator can estimate the impulsive index, Gauss-to-impulsive power ratio, noise power, and hidden states of class A noise model for the channel simultaneously. The considered estimator is different from traditional estimator, which provides a novel method with low-complexity, global optimization capability and potential for parallel processing. Simulation with small sample sizes shows effectiveness of the technique.

Key words: non-Gaussian noise, class A noise, impulsive noise