论文部分内容阅读
提出了一种基于改进误差反向传播神经网络(IBPNN)的具有记忆效应功率放大器(PA)的行为模型。该模型在传统误差反向传播神经网络(BPNN)的基础上利用Levenberg-Marquardt(LM)学习算法和加入动量因子的训练算法更新BPNN的权值和阈值,与传统的BPNN相比只需要更少的训练次数就达到了更高的精度。20MHz带宽三载波WCDMA信号的时域和频域仿真都表明其具有良好的性能,并且由得到的功率放大器(PA)动态特性AM/AM和AM/PM可知,该模型可以很好地描述PA的记忆效应。最后,用16QAM调制的OFDM 20MHz带宽信号的实验证明了该模型具有普遍的适用性。
A behavioral model with memory effect power amplifier (PA) based on improved error back propagation neural network (IBPNN) is proposed. The model updates the weights and thresholds of BPNN using the Levenberg-Marquardt (LM) learning algorithm and the addition of momentum factor training algorithm based on the traditional error backpropagation neural network (BPNN), which only requires less compared with the traditional BPNN The number of training to achieve a higher accuracy. Both the time-domain and frequency-domain simulations of the 20MHz bandwidth three-carrier WCDMA signal show good performance, and the obtained PA power dynamics (AM / AM and AM / PM) show that this model can describe PA Memory effect. Finally, the experiment of OFDM 20MHz bandwidth signal modulated by 16QAM proves that this model has universal applicability.