一种基于随机软反馈Hopfield神经网络的降低OFDM系统峰均比新方案

A New Approach to Reduce PAR by Using S-HNN for OFDM System

该文提出了一种用自适应随机软反馈Hopfield神经网络来优化处理OFDM系统中峰均比(PAR)问题的新方案。通过采用一种较简单的可变动态范围的输出函数,网络的实现复杂度被降低;通过在神经元状态方程加随机扰动的方法来使神经网络能够搜索到最优的相位序列,仿真结果表明这种方法与基于传统HNN的方法相比,OFDM系统PAR性能有了极大地提高,是一种行之有效的实用方案。

This paper proposes a new approach to reduce the Peak-to-Average power Ratio（PAR） value of multi-carrier/OFDM with a new kind of adaptive soft feedback Stochastic Hopfield Neural Network （S-HNN）. By adopting new neural output function and random state disturbance, the system performance is improved greatly. Furthermore, it can be implemented easily compared to the traditional HNN method. By parameter adjustment, the lowest PAR can be found. So it is an effective and practical algorithm for OFDM system.