基于复数径向基函数神经网络的自适应均衡器

Adaptive Equalizer Based on Complex Radial Basis Functions Networks

本文对基于复数径向基函数网络（CRBFN）的自适应均衡器作了研究；提出了两种新的复数径向基函数网络结构的均衡器：正交CRBFN（OCRBFN）均衡器与非对称CRBFN（ACRBFN）均衡器；并在此基础上提出了这两种自适应均衡器的训练算法．模拟结果表明对于线性信道或非线性信道这两种基于CRBFN均衡器的收敛性能比文献［2，5」中的均衡器优越，具有一定的实用价值．

In this pape,the equalizrs based on complex radial basis function networks(CRBFN)were invcestigated and two new adaptive equalizers based on CRBFN(which are orthogonal CRBFN equalizer and asymmetric CRBFN equalizer) were also proposed which are different from those in Refs. [2,5]. Based on both the new CRBFNs, the training algorithm that adapts all be parameters of the network was presented. Computer simulation results show both the new CRBFN equalizers have better performanve than those in Refs. [2,5] when the channel is linear or nonlinear.