正在加载图片...
第24卷第4期 数据采。集与处理 Vol.24 No.4 2009年7月 Journal of Data Acquisition Processing Jul.2009 文章编号:1004-9037(2009)04-0418-05 基于MCMC的CDMA系统联合激活用户识别和信道估计 陈亮辉胡捍英 (解放军信息工程大学通信工程系,郑州,450002) 摘要:用户澈活参数和信道参数彩响了CDMA系统多用户检测算法的性能。根据CDMA接收信号表达式,把接 收估号看作以用户激活数和用户路径数为参数的被来声污染的复合Po550n过程,利用信道脉冲响应硫状特性, 把信道脉冲响应建模为Bernoulli-Gaussian过程,然后使用马尔科夫-蒙特卡罗仿真枝术求出参数后验概率分布 最优化解,实现联合激活用户识别和信道参数估计。由于考虑了匹配滤波后增强的噪声相关性,仿真姑果表明, 在相同信噪比下基于马尔科夫-蒙特卡罗仿真算法估计误差性能优于期望最大化算法和造代条件樸算法· 关键词:复合Poisson过程;澈活用户识别;信道估计,MCMC算法 中围分类号:TN911.72 文献标识码:A Joint Active User Identification and Channel Estimation Using MCMC Methods in CDMA System Chen Lianghui,Hu Hanying (Department of Communication Engineering,PLA Information Engineering University,Zhengzhou,450002,China) Abstract:The performance of multi-user detection (MUD)algorithm in CDMA system is af- fected by parameters of the active user and the channel.The received signal can be viewed as a noise polluted complex Poisson process with parameters of the active user and their paths.By modeling the channel pulse response as a Bernoulli-Gaussian process based on the character of sparse,the optimization of the posterior distribution by Markov Chain Monte Carlo(MCMC) simulation techniques is derived,and the joint active user identification and the channel estima- tion are achieved.Considering the enhanced noise correlation after match filtering,simulating results show that under the condition of the same SNR,the performance of MCMC simulation algorithm is better than that of expectation-maximization (EM)and the iterated conditional mode (ICM)algorithms. Key words:complex Poisson process;active user identification;channel estimation;MCMC al- gorithm (Akaike information criterion,AIC)或最小描述距 分 言 离(Minimum description length,MDL))准则完 成,但是当系统用户激活个数和激活用户的存在路 在异步多用户多径CDMA环境下,用于信道 径幅度变化很大时,算法估计参数的误差很大。通 估计的算法有比如基于期望最大化(Expectation 过匹配技术选代抽取信道路径算法在路径非常近 maximization,EM)最大似然估计算法们和迭代条 时可能导致错误。为了克服这些问题,采用代价函 件模(Iterated conditional mode,ICM)a]等,这些 数算法,其中代价函数项表征了径幅度的先验知 算法需要知道激活用户参数或路径数的先验估计。 识,L,-norm代价函数简单描述了信道的疏状特性 虽然参数的先验估计可以由Akaike信息准则 和参数估计算法),然而L,-norm代价准则的贝叶 基金项日:国防预研基金(40901010201)资助项目, 收稿日期:2008-03-25:修订日期:2008-07-15 万方数据第24卷第4期 数 据 采 集 与 处 理 V01.24 No.4 2009年7月 Journal of Data Acquisition&Processing Jul.2009 文章编号:1004—9037(2009)04.0418.05 基于MCMC的CDMA系统联合激活用户识别和信道估计 引 陈亮辉 胡捍英 (解放军信息工程大学通信工程系,郑州,450002) 摘要:用户激活参数和信道参数影响了CDMA系统多用户检测算法的性能。根据CDMA接收信号表达式,把接 收信号看作以用户激活敷和用户路径数为参数的被噪声污染的复合Poisson过程,利用信道脉冲响应疏状特性, 把信道脉冲响应建模为Bernoulli—Gaussian过程,然后使用马尔科夫-蒙特卡罗仿真技术求出参数后验概率分布 最优化解,实现联合激活用户识剐和信道参数估计。由于考虑了匹配滤波后增强的噪声相关性,仿真结果表明, 在相同信噪比下基于马尔科夫一蒙特卡罗仿真算法估计误差性能优于期望最大化算法和迭代条件模算法。 关键词:复合Poisson过程;激活用户识别;信道估计;MCMC算法 中图分类号:TN911.72 文献标识码:A Joint Active User Identification and Channel Estimation Using MCMC Methods in CDMA System Chen Lianghui,Hu Hanying (Department of Communication Engineering。PLA Information Engineering University,Zhengzhou,450002,China) Abstract:The performance of multi—user detection(MUD)algorithm in CDMA system is af— feeted by parameters of the active user and the channel.The received signal can be viewed as a noise polluted complex Poisson process with parameters of the active user and their paths.By modeling the channel pulse response as a Bernoulli—Gaussian process based on the character of sparse,the optimization of the posterior distribution by Markov Chain Monte Carlo(MCMC) simulation techniques is derived,and the joint active user identification and the channel estima— tion are achieved.Considering the enhanced noise correlation after match filtering,simulating results show that under the condition of the same SNR,the performance of MCMC simulation algorithm iS better than that of expectation—maximization(EM)and the iterated conditional mode(ICM)algorithms. Key words:complex Poisson process;active user identification;channel estimation;MCMC al— gorithm 1=1 在异步多用户多径CDMA环境下,用于信道 估计的算法有比如基于期望最大化(Expectation maximization,EM)最大似然估计算法‘13和迭代条 件模(Iterated conditional mode,ICM)[2]等,这些 算法需要知道激活用户参数或路径数的先验估计。 虽然参数的先验估计可以由Akaike信息准则 基金项目:国防预研基金(40901010201)资助项目。 收稿日期:2008—03—25;修订日期:2008—07—15 (Akaike information criterion,AIC)或最小描述距 离(Minimum description length,MDL)L30准则完 成,但是当系统用户激活个数和激活用户的存在路 径幅度变化很大时,算法估计参数的误差很大。通 过匹配技术迭代抽取信道路径算法在路径非常近 时可能导致错误。为了克服这些问题,采用代价函 数算法,其中代价函数项表征了径幅度的先验知 识,L。一norm代价函数简单描述了信道的疏状特性 和参数估计算法[.],然而L。一norm代价准则的贝叶 万方数据
向下翻页>>
©2008-现在 cucdc.com 高等教育资讯网 版权所有