切比雪夫函数型连接神经网络在信道均衡中的应用

本文提出一种基于切比雪夫函数型连接神经网络(CFLNN)的信道均衡方法。传统的前馈神经网络虽然能有效地解决信道均衡的问题,但具有计算复杂度过高,收敛速度慢等缺点。函数型连接神经网络通过对输入模式进行非线性扩展,可以不必使用隐层而不降低整体性能,从而极大简化了网络结构。同时,神经网络的学习方法得以简化,提高了收敛速度。本文采用可变尺度共扼梯度下降法(SCG)对该函数型连接网络进行训练。仿真结果表明了用切比雪夫函数型连接神经网络解决信道均衡问题的有效性。

The Application of Chebyshev Functional Link Neural Networks in Channel Equalization

A computationally efficient neural network named as chebyshev functional link neural network (CFLNN) for the purpose of channel equalization is proposed in this paper. A channel equalizer can be effectively implemented by using feed forward neural networks, but this also brings the drawback of high computational complexity and slow convergence rate. A functional link neural network can expand its input pattern to eliminate the need of hidden layer without sacrificing its performance. Thus the complexity of network and the learning algorithm can be simplified remarkably with a faster convergence rate. This CFLNN equalizer is trained by a scaled conjugate gradient (SCG) algorithm. The simulation results, indicate that CFLNN is an efficient method to the channel equalization problem.