Legendre神经网络非线性信道均衡

在分析Chebyshev正交多项式神经网络非线性滤波器的基础上，利用Legendre正交多项式快速逼近的优良特性以及判决反馈均衡器的结构特点，提出了两种新型结构的非线性均衡器，并利用NLMS算法，推导出自适应算法．仿真表明，无论通信信道是线性还是非线性，Legendre神经网络自适应均衡器与Chebyshev神经网络均衡器的各项性能均接近，而Legendre神经网络判决反馈自适应均衡器能够更有效地消除码间干扰和非线性干扰，误码性能也得到较好的改善．

Equalization of Nonlinear Channel in Legendre Neural Network

Based on the analysis of Chebyshev orthogonal polynomial neural network nonlinear filter,and by using the fast approximation characteristics of Legendre orthogonal polynomial and the structrue of decision feedback equalizer,two nonlinear equalizers with new structures are proposed,and an adaptive algorithm is deduced with the normalized least mean squares(NLMS).Simulations show that the equalization performances of the adaptive equali-zers based on Legendre neural network and Chebyshev neural network are very approximate no matter the channel is linear or nonlinear,and the adaptive decision feedback equalizer based on Legendre orthogonal polynomial neural network can more effectively remove the nonlinear disturbance and intersymbol interference(ISI) and impove the performance of bit error ratio(BER).