摘要
在一个信道上传送多个用户信号，可以大大提高信道的容量．本文讨论了非理想信道多用户数字信号的盲分离问题．利用天线阵，接收信号可以看作是由Ｎ个独立信号源激励的线性传输混合系统的输出；由于信道存在码间干扰，混合矩阵的元素不是常数，而是一个线性子系统．针对这一情况，本文提出了一个新的盲分离器结构，首先将接收信号进行盲分离，然后利用基于ＡＲ模型的盲均衡器消除每一路输出信号的码间干扰，从而实现多用户信号的分离．文中给出了盲分离器的神经网络学习算法，讨论了其稳定性及收敛性．模拟结果显示分离效果是令人满意的．
Co channel transmission can dramatically increase channel capacity in wireless digital communication.In this paper,a recursive neural network is proposed for separating non ideal cochannel digital signals arriving at an antenna array.The received signals of the antenna array can be thought as the outputs of a linear mixing transmission system driven by N independent sources.Since the channels are not ideal,each element of the mixing array is a linear sub system instead of a constant.To recover the transmitted signals,the recursive neural network is used to realize signal separation and channel equalization.Learning algorithm of the neural network is presented and its stability is discussed.Simulation results demonstrate its good convergence property and anti noise performance in recovering both PAM and QAM signals.
出处
《电子学报》
EI
CAS
CSCD
北大核心
1999年第1期41-44,共4页
Acta Electronica Sinica
关键词
盲分离
多用户数字信号
非理想信道
信道均衡
Blind separation,Co channel digital signal,Non ideal channel,Antenna array
