摘要
针对卷积混合信号,提出了一种新的多通道盲解卷积算法,该算法首先利用子空间分解方法,将信号卷积混合模型变换成线性混合模型,然后利用线性混合盲分离算法分离出源信号。该算法相对频域盲解卷积算法来说无需解决线性混合盲分离中存在的幅度和排列顺序的模糊性问题,而且该算法不要求信号独立同分布,只要求各源信号统计独立即可。因此,该算法可以直接在中频对观察信号进行处理。计算机仿真结果表明,该算法不仅能对不同频不同调制方式的通信信号进行盲解卷积,而且对同频同调制的通信信号,该算法同样有效。
Aim at the blind separation of convolution mixture signals, a novel algorithm for multi-channel blind deconvolution was presented. Firstly, this algorithm transforms the convolution mixture into the instantaneous linear mixture by subspace decomposition. The second step is to separate the estimated linearized signals and to obtain the source signals by using the blind source separation algorithm for instantaneous linear mixture. This algorithm needs not to solve the problem of the scale and permutation ambiguity existing in the instantaneous linear blind separation relative to the frequency domain method for blind deconvohition. Moreover, the algorithm doesn＇t need signal is independent identically distributed（i.i.d）, only needs the sources are statiscally independent each other. Therefore the algorithm can process the mixing signal in intermediate frequency. Simulation results show that this algorithm can blind deconvolute not only the signals with the different carrier frequencies and different symbol rates, but also that with same carder frequencies and same symbol rates.
出处
《通信学报》
EI
CSCD
北大核心
2009年第1期25-30,共6页
Journal on Communications
基金
国家自然科学基金资助项目(60702060)
高等学校学科创新引智计划资助项目(B08038)~~
关键词
盲信号处理
盲解卷积
子空间分解
多通道
blind signal processing
blind deconvolution
subspace decomposition
multi-channel
