基于动态学习率深度神经网络的抗干扰信道编码算法

Anti-interference channel coding algorithm based on dynamic learning rate deep neural network

针对电子战条件下,通信信号易受压制干扰的问题,提出了一种基于动态学习率深度自编码器(dynamic learning rate deep Auto Encoder,DLr-DAE)的信道编码算法来提高系统抗压制干扰性能。首先对输入未编码信号进行预处理,将原始输入信号转换为单热矢量;随后使用训练数据样本集,用非监督学习方法训练深度自编码器,基于随机梯度下降法(SGD)更新网络参数,利用指数衰减函数,在迭代次数和网络损失函数值变化过程中动态微调学习率,减少网络迭代循环次数,避免收敛结果陷入局部最优点,从而获得面向电子战环境的信道编码深度学习网络。仿真结果表明,相比现有深度学习编码算法,该算法在取得同等误码率时,抗噪声压制干扰性能最大可提升0.74 d B。

Aiming at the situation that the communication signal is vulnerable to hostile suppression under the condition of electronic warfare,this paper proposed a new channel coding scheme based on DLr-DAE.It could improve the performance of the communication system against suppression interference.First,it preprocessed and converted the original input signal into a one-hot vector.Then it used the training data sample set to train the deep AutoEncoder in an unsupervised learning method,and updated the network parameters according to the stochastic gradient descent(SGD)method.In this process,it used the exponential decay function to continuously fine-tune the learning rate according to the number of iterations and the value of the network loss function.By this method,it reduced the network optimization epoch and avoided the convergence result to the suboptimal point.Thus,it obtained a channel coding deep learning network for the electronic warfare environment.The simulation results show that the proposed algorithm can improve the anti-suppression noise interference performance by 0.74 d B,compared with the traditional deep learning coding algorithm when obtained the same bit error rate.