Experimental Design of Deep Neural Network Channel Estimation Simulation for Orthogonal Frequency Division Multiplexing Underwater Acoustic Communication
For orthogonal frequency division multiplexing(OFDM)underwater acoustic communication system,the least squares(LS)channel estimation algorithm is greatly affected by noise,and the minimum mean square error(MMSE)algorithm needs the prior statistics of the channel and the calculation is large.To solve these problems,a channel estimation algorithm based on deep neural network(DNN)is proposed.First,the DNN model is constructed at the receiving end,and then the Bellhop underwater acoustic channel ray model is used to complete the training,verification,testing and correction of the DNN model.Finally,the working scene simulation experiment is designed with the OFDM system,and the comparison with the traditional channel estimation algorithm is conducted.The experimental results show that when the bit error rate is less than 10-4,DNN algorithm can reduce 2 dB compared with LS algorithm and 1.5 dB compared with MMSE algorithm.The new method proposed will optimize and integrate the channel estimation related problems of traditional OFDM underwater acoustic communication system.The channel estimation does not require mass pilot data and channel prior statistics information,and the system can establish nonlinear mapping in time,with advantages of less samples and fast convergence,which is of great significance for saving system storage resources and improving real-time performance.
Orthogonal Frequency Division MultiplexingUnderwater Acoustic CommunicationDeep Neural NetworkChannel Estimation
万方数据
维普
