Deep neural network based on polar code for improving the BER performance of the non-Gaussian underwater acoustic communication system

Abstract

The neural network is one of the techniques used to enhance the bit error rate performance in communication systems. The wireless communication system usually follows the standard Gaussian distribution, while the noise in underwater acoustic systems usually follows non-Gaussian distribution and leads the system to a high bit error rate value. In this paper, the deep forward neural network has been adopted to improve the bit error rate of the underwater aquatic system based on t-distribution. Furthermore, the parameters of the deep neural network, number of nodes, number of layers, and activation functions have been evaluated to note the behavior of the bit error rate performance of the system. The results show that the bit error rate value of the system declines when increasing the number of nodes and layers in the neural network, while the tanh activation function is a suitable function that can be used to improve the bit error rate performance of the system. Moreover, the complexity and latency pattern of the system increases by boosting the number of nodes and layers of the neural network.