基于LEAP神经网络同步DS-CDMA伪码序列盲估计
发布时间：2018-08-15 19:26
【摘要】：针对特征分解方法在实现非等功率同步直接序列码分多址(DS-CDMA)信号伪码序列盲估计时存在的处理数据向量不能太长以及不能工作于非平稳环境中的问题,引入了一种由主分量分析实现自适应特征提取的在线无监督学习(LEAP)神经网络(NN)。首先将已分段的一周期DS-CDMA信号作为NN的输入信号,用NN各权值向量的符号函数代表DS-CDMA信号各用户的伪码序列,然后通过不断输入信号来反复训练权值向量直至收敛,最终DS-CDMA信号各用户的伪码序列就可以通过各权值向量的符号函数重建出来。此外,采用变步长以提高收敛速度。理论分析与仿真实验表明,LEAP NN至少可以实现-20 d B信噪比下10个用户的非等功率同步DS-CDMA伪码序列盲估计,并且比传统的Sanger NN具有更快的收敛速度。
[Abstract]:Aiming at the problem that the processing data vector can not be too long and can not work in non-stationary environment when the eigen decomposition method realizes blind estimation of pseudo-code sequence of non-equal power synchronous direct sequence code division multiple access (DS-CDMA) signals, an on-line unsupervised learning (LEAP) neural network based on principal component analysis (PCA) for adaptive feature extraction is introduced. Network (NN). Firstly, the segmented one-cycle DS-CDMA signal is used as the input signal of NN, and the symbolic functions of the weighted vectors of NN are used to represent the pseudo-code sequences of the users of DS-CDMA signal. Then the weighted vectors are trained repeatedly until convergence through continuous input signals. Finally, the pseudo-code sequences of the users of DS-CDMA signal can pass through each weighted direction. In addition, the variable step size is used to improve the convergence rate. Theoretical analysis and simulation results show that LEAP NN can achieve at least 10 users'non-equal-power synchronous DS-CDMA pseudo-code sequence blind estimation under -20 D B SNR, and has faster convergence rate than the traditional Sanger NN.
【作者单位】： 重庆邮电大学信号与信息处理重庆市重点实验室;
【基金】：国家自然科学基金资助项目(61371164,61275099) 信号与信息处理重庆市市级重点实验室建设项目(CSTC2009CA2003) 重庆市杰出青年基金项目(CSTC2011jjjq40002) 重庆市教育委员会科研项目(KJ130524) 重庆市研究生科研创新项目(CYS14140)
【分类号】：TP183;TN911.23
本文编号：2185182
[Abstract]:Aiming at the problem that the processing data vector can not be too long and can not work in non-stationary environment when the eigen decomposition method realizes blind estimation of pseudo-code sequence of non-equal power synchronous direct sequence code division multiple access (DS-CDMA) signals, an on-line unsupervised learning (LEAP) neural network based on principal component analysis (PCA) for adaptive feature extraction is introduced. Network (NN). Firstly, the segmented one-cycle DS-CDMA signal is used as the input signal of NN, and the symbolic functions of the weighted vectors of NN are used to represent the pseudo-code sequences of the users of DS-CDMA signal. Then the weighted vectors are trained repeatedly until convergence through continuous input signals. Finally, the pseudo-code sequences of the users of DS-CDMA signal can pass through each weighted direction. In addition, the variable step size is used to improve the convergence rate. Theoretical analysis and simulation results show that LEAP NN can achieve at least 10 users'non-equal-power synchronous DS-CDMA pseudo-code sequence blind estimation under -20 D B SNR, and has faster convergence rate than the traditional Sanger NN.
【作者单位】： 重庆邮电大学信号与信息处理重庆市重点实验室;
【基金】：国家自然科学基金资助项目(61371164,61275099) 信号与信息处理重庆市市级重点实验室建设项目(CSTC2009CA2003) 重庆市杰出青年基金项目(CSTC2011jjjq40002) 重庆市教育委员会科研项目(KJ130524) 重庆市研究生科研创新项目(CYS14140)
【分类号】：TP183;TN911.23
【相似文献】
相关期刊论文 前7条
1 陈实如,张京娟,孙尧;直扩系统伪码序列串并组合捕获方案[J];哈尔滨工程大学学报;2003年01期
2 徐颖;封欣;吴嗣亮;;基于FPGA的任意时延伪码序列产生方法[J];微计算机信息;2009年08期
3 金红军;战术跳频电台伪码序列的研究[J];电讯技术;1998年06期
4 李军伟;张天骐;朱洪波;潘毅;;基于聚类的多进制扩频伪码序列盲估计方法[J];科学技术与工程;2014年02期
5 韦永朋;刘洛琨;郭虹;杨正举;;一种多径信道直扩信号伪码序列盲估计方法[J];计算机工程与应用;2013年12期
6 沈斌;王建新;;基于奇异值分解的直扩信号伪码序列及信息序列盲估计方法[J];电子与信息学报;2014年09期
7 ;[J];;年期
，本文编号：2185182
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2185182.html
