论文部分内容阅读
针对超奈奎斯特速率传输信号在传输过程中产生的严重码间干扰问题,提出了一种基于卷积神经网络(CNN)的解调器,对双极性扩展的二进制相移键控(bipolar EBPSK)超奈奎斯特速率信号进行解调.利用卷积神经网络局部感受野、池化和权值共享的特点,提出了一种具有6层结构的卷积神经网络来解调扩展的二进制相移键控调制信号并消除码间干扰.实验结果表明:当码率为1.07 k Bd、发送端带宽限制为1 k Hz,且一个码元中跳变载波周期数K=5,10,15,28时,CNN单码元判决方法误码率性能总体优于CNN双码元联合判决方法;当K等于码元载波周期总数N,即K=N=28时,CNN单码元判决误码率方法优于相干解调约0.5 d B;当码率为1.07 k Bd、发送端带宽限制为500 Hz,且K=5,10,15,28时,CNN双码元联合判决方法优于CNN码元判决方法;当K=N=28时,CNN双码元判决方法优于相干解调约0.5~1.5 d B.基于CNN的解调器成功地解决了由超奈奎斯特速率双极性传输信号产生的严重码间干扰问题,有利于频谱利用率的提高.
Aiming at the serious inter-symbol interference problem of super-Nyquist rate transmission signal during transmission, a convolution neural network (CNN) -based demodulator is proposed for bipolar extended binary phase shift keying bipolar EBPSK) demodulation.According to the characteristics of local receptivity, pooling and weight sharing of convolutional neural networks, a convolutional neural network with 6-layer structure is proposed to demodulate the extended Binary phase shift keying modulation signal and eliminate intersymbol interference.The experimental results show that when the code rate is 1.07 k Bd and the transmit end bandwidth is limited to 1 k Hz and the number of hopping carrier periods K = 5 and 10 in one symbol, 15, 28, the BER performance of CNN single-symbol decision method is better than that of CNN dual-symbol joint decision method. When K equals the total number of symbol carrier cycles N, ie, K = N = 28, The code rate method is better than the coherent demodulation for about 0.5 dB. When the code rate is 1.07 kBd and the transmitter bandwidth is limited to 500 Hz, and the K = 5, 10, 15, and 28, the CNN dual symbol joint decision method is superior to CNN symbol decision method; when K = N = 28, CNN dual symbol decision method is better than coherent demodulation about 0.5 ~ 1.5 d B. CNN-based demodulator successfully solved The problem of serious inter-symbol interference caused by super-Nyquist rate bipolar transmission signals is solved, which is favorable for improving the spectrum utilization rate.