摘要
针对非合作接收PCMA混合信号盲分离中高复杂度束缚,提出一种基于前馈神经网络的分离算法,通过搭建神经网络分离平台,规避传统的发送符号遍历思想,实现PCMA混合信号低复杂度高性能盲分离.仿真实验表明,神经网络能够极大挖掘信号内在信息,针对QPSK调制PCMA混合信号,在信噪比7dB时误比特率达到10^(-3)数量级,并伴随着较PSP分离算法算术平方根级别的复杂度降低.
Aiming at the high complexity in blind separation of PCMA mixed signals with non-cooperative reception,the separation algorithm based on feedforward neural network is proposed.By setting up a neural network separation platform and avoiding the traditional idea of maximum a posteriori probability,the blind separation algorithm with low complexity and high performance can be realized.Simulation results show that the neural network can greatly exploit the intrinsic information of the signal,and 10 -3 orders of bit error rate performance is achieved with 7 dB of signal-to-noise ratio to QPSK modulated PCMA signals,accompanied by the declining complexity of the arithmetic square root level compared with the PSP algorithm.
作者
郭一鸣
彭华
杨勇
GUO Yi-ming;PENG Hua;YANG Yong(PLA Information Engineering University,Zhengzhou,Henan 450002,China;61886 Troops of PLA,Beijing 100084,China)
出处
《电子学报》
EI
CAS
CSCD
北大核心
2019年第2期302-307,共6页
Acta Electronica Sinica
基金
国家自然科学基金(No.61401511
No.U1736107)
关键词
神经网络
非合作
成对载波多址复用
盲分离
neural network
non-cooperative
Paired Carrier Multiple Access (PCMA)
blind separation
