基于LMK准则的前馈神经网络盲多用户检测算法  

Blind Multi-user Detection Algorithm of Feed-forward Neural Network Based on LMK Criteria

作　　者：孙云山[1] 李昌华[1] 马学伟[1] 杨淼[1] 方英杰[1] 谢蓓雷[1] 杨艺[1] 

机构地区：[1]天津商业大学信息工程学院,天津300134

出　　处：《天津商业大学学报》2008年第3期17-20,共4页Journal of Tianjin University of Commerce

摘　　要：提出了一种基于前馈神经网络的盲多用户检测算法,该算法基于最小平均峰度准则构造了一个代价函数,给出了算法的约束条件,利用罚函数方法对带约束的代价函数进行优化,获得前馈神经网络网络权值的迭代公式,从而实现盲多用户检测。计算机仿真结果表明,新算法具有较强的多址干扰抑制性能。A novel blind multi-user detection algorithm based on feed-forward neural network （FNN） is proposed in the paper. A cost function based on Least Mean Kurtosis（LMK） criterion is founded, and the restraint condition of cost function is deduced. Weights and parameters of feed-forward neural network are optimized by punishment function, iteratire formulas of FNN are obtained, and then blind multi-user detection algorithm is realized. Simulation results indicate that the novel algorithm has lower bit-error ratio and better convergence ability.

关 键 词：盲多用户检测 前馈神经网络 LMK准则 

分 类 号：TN911.5[电子电信—通信与信息系统] TP183[电子电信—信息与通信工程]