改进的暂态混沌神经网络多用户检测器  

An improved multi-user detector for transiently chaotic neural networks

作　　者：谢红[1] 刘瑶[1] 

机构地区：[1]哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001

出　　处：《应用科技》2008年第6期14-17,共4页Applied Science and Technology

摘　　要：提出了一种基于自反馈连接权值优化策略的暂态混沌神经网络多用户检测方法.该方法采用一种改进的变指数退火函数,对自反馈权值进行优化,既充分利用混沌的动态特性进行搜索,使算法可以从局部最优值中跳出,又加快收敛速度,减少收敛时间.仿真结果表明,这种检测器在误码率和抗远近效应方面优于已有的神经网络多用户检测器,能较好地逼近最佳多用户检测器的性能.A multi-user detector for transiently chaotic neural network （ TCNN ） based on optimized value of the selffeedback connection weights was proposed. This method optimizes the self-feedback connection weights by using an improved variable-exponent annealing function, because it uses the chaotic dynamic property to carry out the searching process, skipping the local optimum, accelerating the searching speed and reducing the convergence time. The simulation results showed that the bit error rate and near-far resistance of the proposed detector are obviously superior to existing neural network detectors and the detector is comparable with the optimal detector.

关 键 词：多用户检测 暂态混沌神经网络 自反馈连接权值 

分 类 号：TN914.53[电子电信—通信与信息系统]