惩罚函数优化的前馈神经网络盲多用户检测  

Feed-forward Neural Network Blind Multi-user Detection by Penalty Function Optimization

作　　者：孙云山[1] 张立毅[1] 刘婷[1] 李艳琴[1] 

机构地区：[1]天津商学院信息工程学院,天津300134

出　　处：《计算机工程》2008年第11期194-196,共3页Computer Engineering

基　　金：中国博士后科学基金资助项目(20060390170);山西省自然科学基金资助项目(20051038)

摘　　要：提出一种前馈神经网络盲多用户检测算法,利用前馈神经网络替代原有检测器中的滤波器,通过惩罚函数对约束恒模代价函数进行求解,获得前馈神经网络权值和参数的迭代公式,实现了盲多用户检测。Matlab仿真结果表明,该算法改善了系统的误码率性能,加快了算法的收敛速度。A blind multi-user detection algorithm based on feed-forward neural network is proposed. A feed-forward neural network replaces the fiker of the original detector. A constrained constant modulus algorithm cost function is obtained by a penalty function. Iterative formula of weights and parameters of neural networks are acquired. The blind multi-user detection algorithm is realized. MATLAB simulation indicates that the new algorithm improves system performance such as bit error rate and convergence speed.

关 键 词：盲多用户检测 前馈神经网络 恒模算法 惩罚函数 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]