基于神经网络的无线通信系统RF电路设计方法

典型的无线通信系统RF电路由大量的线性和非线性元器件组成。建立这些微波电路元器件的模型在现代RF电路的设计中至关重要。采用一种神经网络来学习非线性电路的动态特性,并建立相应的模型。神经网络模型可以对原始电路的模拟特性进行整体预测,可用于高层电路的仿真和优化。以功率放大器为例验证了神经网络模型的实用性,表明神经网络模型可以极大地提高无线通信系统RF电路的建模精度和速度。

Design Approach for Wireless Communication System RF Circuit Based on Neural Networks

Typical wireless communication system RF circuits comprise a large number of linear and non-linear components.In this paper,we employ a neural network(NN) to learn the dynamic characteristics of nonlinear circuits and establish their model.The NN model provides integral prediction of the full analog behavior of the original circuit and can be useful for high-level simulation and optimization.An practical example of modeling a power amplifier is used to demonstrate the utility of the proposed modeling approach.It shows that we can speed up the modeling RF circuits by using the neural network approach considerably.