基于RBF网络信道均衡器

在数字通信系统中为了克服信道畸变引起的码间干扰，在接收端必须采用信道均衡技术。在本文中，我们将RBF网络用作均衡器。采用最近邻聚类和直接判决算法来调整隐藏层中心，然后再用LMS算法调整输出层的系数。该算法可以实现在线学习，根据相应的准则增加，删除隐藏层节点。算法事先不必确定隐藏层的节点个数。模拟结果显示，RBF网络均衡器能够正确地将信号从有噪信道中恢复出来，在计算机模拟仿真中其性能与理想贝叶斯均衡器相当。

Study on Channel Equalization Based on RBF Networks

High-speed communication channels are often impaired by channel inter-symbol interference and additive noise.Adaptive equalizers are required in these communication systems to obtain reliable data transmission.In this paper,we use RBF networks for channel equalization.The RBF networks' learning algorithm is usually decoupled in two steps.First,clustering is performed on the input vectors to determine the unit centers,and then the output weights are adjusted.In our works,the first step is accomplished using the unsupervised nearest clustering algorithm(NNCA) or the decision-directed algorithm and then the output weights are adjusted using the supervised least mean square(LMS) algorithm.This algorithm uses online learning,and has the capability to grow and prune the RBF network's hidden neurons ensuring a parsimonious network structure.The proposed method eliminates the need for channel order estimation for simulation,Results showing that the RBF network equalizer has the ability to correctly classify the transmitted data form the noise corrupted channel output.The performance of the RBF network and the Bayesian equalizer are seen to be indistinguishable.