基于SCFNN之PAM非线性信道均衡器成效研究

A SCFNN Based PAM Channel Equalizer Performance Research

摘要:

自组织型模糊类神经网络(SCFNN)可依据一定的法则自我构建神经网络的组织结构,从而适用于当前控制对象；多层神经元是传统的类神经网络,广泛应用于各个领域；倒传递学习法与最陡坡降法相结合,可使以上两种类神经网络进行有效的融合；目前,信道均衡器上的系统架构种类非常多,各种类神经网络应用于信道均衡器也颇为普遍；在研究SCFNN的基础上,将其应用于通道均衡器确实可行,效果良好；比较了SCFNN与MLP在通道均衡器的成效；仿真表明,在相同通道环境下,SCFNN的训练收敛速度、位错误率与系统敏感度优于MLP,完成结构学习后SCFNN的结构也颇为精简。

Abstract:

Self-Constructing Fuzzy Neural Network(SCFNN)can create a fuzzy neural network for a target in accordance with a dedicated algorithm.Multi-layer Perceptron(MLP)neural network is a very traditional neural network and many applications were developed in different fields.Back Propagation(BP) combined with steepest descent method make the SCFNN and MLP learned efficiently.Today, many kinds of channel equalizers were constructed, and many kinds channel equalizers based on neural network were also constructed.We prove that the SCFNN can be a superior equalizer. We also compare the performance of SCFNN and MLP applied in channel equalizer. The simulation results show the SCFNN is superior than the MLP in convergence speed, bit error rate and sensitivity. When the SCFNN learning processes is completed the, we found the structure is very simpler.