联合串行干扰抵消与因子图的单通道混合信号盲分离算法  

A Blind Separation Algorithm of Single-Channel Signals Based on Serial Interference Cancellation and Factor Graphs

作　　者：郭一鸣[1] 彭华[1] 张冬玲 杨勇[1] 

机构地区：[1]解放军信息工程大学信息系统工程学院,郑州450001

出　　处：《西安交通大学学报》2015年第10期130-135,共6页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金资助项目(61401511)

摘　　要：为解决复杂编码方式下单通道同频混合信号盲分离复杂度高的问题,提出了一种联合串行干扰抵消(SIC)与因子图的迭代分离算法。首先由传输信道参数来重构发送信道,然后根据重构信道进行SIC解调,获得解调信号软信息传递给译码模块,将译码增益反馈回SIC解调模块,利用迭代处理来实现分离。重点研究了解调与译码间信息迭代交互过程,以及2路信号分量的信道参数差异对分离性能的影响,得到随着迭代次数的增加以及信道参数差异性增加,算法分离性能将得到提升的结论。仿真实验表明,在LDPC编译码环境下,QPSK调制混合信号经2次迭代处理,分离性能较M-PSP算法提升2dB以上。An iteration processing algorithm based on serial interference cancellation（SIC） and factor graph decoding is proposed to solve the high complexity problem of blind separation of single-channel co-frequency signals in complex coding. The forward channels are reconstructed based on the transmission channel parameters. Then, the SIC demodulation is completed under the premise of the channel reconstruction to obtain the soft information and to pass the information to the decoding module. The separation performance is improved with decoding gain feedback. Focuses are laid on the information interaction between demodulation and decoding, and the impact of the difference between channel parameters of signal components of two channels on the performance of demodulation. It is concluded that the separation performance of the algorithm is improved when the number of iterations increases and the difference of channel parameters increases. Simulation results and a comparison with the M-PSP algorithm with LDPC codes show that the proposed algorithm obtains a gain of about 2 dB in signal-noise ratio for QPSK signals after two iterations.

关 键 词：盲分离 单通道 串行干扰抵消 因子图 

分 类 号：TN911.7[电子电信—通信与信息系统]