连续多阈值神经元反馈神经网络盲检测光基带信号

Blind Optical Baseband Signals Detection Using Recurrent Neural Network Based on Continuous Multi-Valued Neurons

空间分集接收可补偿信道衰落,提出了一种基于幅相联合激励法的连续多阈值神经元反馈神经网络(RNNCMVN)的光基带信号直接盲检测方法。针对多进制相移键控(MPSK)信号的特点,设计了两种连续相位多阈值激励函数形式,并简要讨论了该两类激励函数参数的选择;分别推演基于幅相联合激励法的RNNCMVN光基带信号盲检测方法工作于同步和异步两种模式下的新能量函数及其相关证明。同时针对正交调幅(QAM)信号的特点,分别设计出连续振幅和相位多阈值激励函数形式,最后通过仿真验证了该方法的有效性。

Spatial diversity reception can be used over wireless optical links to resist channel fading. A novel blind optical baseband signals detection direct algorithm is proposed by using recurrent neural network based on continuous multi-valued neurons （RNNCMVN） with amplitude and phase continuous activation （APCA）. Considering the characteristics of mary phase shift keying （MPSK） signals, two types of continuous multi-valued activation functions are designed and the method of selecting their parameters is illustrated briefly. The new energy functions of synchronous and asynchronous mode of the RNNCMVN are derived and proved, respectively. Considering the characteristics of quadrature amplitude modulation （QAM） signals, the new continuous amplitude and phase multivalued activation functions are designed and discussed, respectively. Finally, simulation results demonstrate the effectiveness of the proposed method.