空间电子探测信号盲分离研究

给出了盲信号分离中的瞬时混合,时延混合和卷积混合三种混合模型,介绍了两种具体的盲分离算法,等变自适应盲分离算法和非高斯性最大化的快速定点算法.其中对于窄带源信号,对时延混合模型进行了扩展,提出了用复数域瞬时盲信号分离算法分离时延混合信号的新思路.最后给出了相应的仿真和实验结论,实验结果表明用基于复数的盲分离算法确实能够有效地分离阵列接收的时延混合信号.

Research on blind signal separation for spatial electronic detection

First, three kinds of mixture models in BSS are presented, such as instantaneous mixture model, time-delay mixture and convolutive mixture model. Then, two kinds of BSS algorithms are discussed, Equivalent Adaptive Separation via Independence and Fast Fixed-Point Algorithm based on maximization of nongaussianity. For the narrow band signals, the time-delay mixture model is expanded. And a new idea about separating the time-delay mixture signals with the instantaneous BSS algorithms for complex-valued signals is presented. Finally the corresponding simulation results and theoretical test conclusion are carried out. From the test results, the BSS algorithms for complex-valued signals can separate the received array time-delay mixture signals efficiently.