LDPC-BICM图模型中新的非参数消息调度机制  

A Novel Nonparametric Messages Update Schedule in Graphical Model of LDPC-BICM

作　　者：段琳琳[1,2] 王忠勇[1] 高向川[1] 肖岩[1] 

机构地区：[1]郑州大学信息工程学院,郑州450001 [2]解放军信息工程大学信息系统工程学院,郑州450001

出　　处：《信号处理》2014年第9期1000-1006,共7页Journal of Signal Processing

基　　金：国家自然科学基金(61172086;61201251);国家自然科学基金联合基金(U1204607);博士后科研启动基金(2011012)

摘　　要：针对时变信道下LDPC编码的比特交织编码调制(Low Density Parity-Check coded Bit-Interleaved Coded Modulation,LDPC-BICM)系统的因子图模型中信道估计精度低与复杂度较高等问题,改进传统参数消息调度机制,提出一种新的非参数消息调度机制实现联合信道估计与译码。为获得性能与复杂度的折中,一方面从全局角度考虑,合理选择不同的内外迭代组合形式,设计了稀疏估计的消息调度机制,一方面从迭代译码角度考虑,通过选择不同的门限和比较准则,设计了局部选择性更新的消息调度机制。同时,将max-sum规则与MCMC粒子滤波相结合,给出基于硬判决的信道消息更新机制,有效解决了粒子样本枯竭问题,提高了信道估计精度。仿真结果表明,这种消息调度机制不仅有效降低复杂度,并且优于传统的消息调度机制。Compared to conventional message schedule in factor graph model of LDPC-BICM （ low density parity-check coded Bit-Interleaved Coded Modulation） system, a novel nonparametric message update schedule is proposed to get a good trade-off between the performance and complexity. From a global perspective, a message update schedule with sparse channel estimation is designed by selecting different kinds of combination of the outer and inner iteration. And a partial selective message update schedule is designed from the local iterative decoding perspective by choosing proper comparison rule and threshold. In addition, a channel message update schedule is presented to prevent particle impoverishment and improve channel estimation precision by combining Markov Chain Monte Carlo （MCMC） particle filters （PF） into max-sum rule. Simulation results illustrate effectiveness and performance superiority of the proposed schedule compared to conventional schedule.

关 键 词：LDPC编码的比特交织编码调制 因子图 最大和规则粒子滤波 消息调度机制 

分 类 号：TN911.23[电子电信—通信与信息系统]