Design Of Maritime Communication System Based On Autoencoder

With the rapid development of maritime economy,the requirements for stable maritime communication technology are getting.Due to the more harsh marine environment which is affected by severe weather such as wind,rain,thunder and lightning,there are limited options in existing marine communication methods.Therefore,how to improve the throughput and reliability of maritime communication at the physical layer is crucial to the development of the Internet of Ships and the Internet of Things at sea.According to the analysis of the characteristics of maritime wireless communication,this thesis studies the end-to-end maritime autoencoder based on Convolutional Neural Network(CNN),aiming to establish a maritime wireless communication system with high reliability and high throughput.In order to design a reliable maritime wireless communication system,it is necessary to analyze the characteristics of maritime wireless channels and determine a channel model suitable for maritime communication firstly.Because the marine environment has fewer obstacles than land,and the ocean area is relatively vast,there will be a large number of direct signals.On the basis of the research and analysis of a large number of literatures,this thesis determines the Rician fading channel as the channel model of the maritime communication system.In order to improve the reliability and the throughput of maritime communication system,this thesis designs a CNN-based end-to-end maritime autoencoder communication system(CNN-MAE).Compared with the traditional maritime communication system,which can only achieve the optimal design of a single module,CNN-MAE can achieve the overall optimal performance of the entire maritime communication system.Meanwhile,CNN-MAE optimizes all modules at the same time,so as to achieve the optimal performance of the system and further improve the reliability of the communication system.At the same time,the CNN-MAE communication system can improve the generalization ability of the system and eliminate the floor effect.The simulation results show that the bit error performance will still decrease in the case of a large signal-to-noise ratio.In order further improve the reliability of the communication system further,this thesis proposes an autoencoder communication system based on feedback mechanism,namely FBCCNN-MAE system.The system adds a feedback decoder module and a channel feedback module at the transmitting end and the channel respectively.The part of feedback decoder is implemented by the CNN scheme,which solves the defect that the feedback channel is too much affected by noise in the traditional feedback encoder.In addition,FBC-CNN-MAE adopts the method of partial feedback,applying interval sampling in the feedback process to realize the extraction of information,and uses the feedback decoder to realize the feedback signal reduction and extraction,so as to reduce the transmission power of the system on the basis of improving the reliability of the system.In this thesis,in order to cope with the harsh communication environment at sea and build a highly reliable and high-throughput maritime wireless communication system,the CNN-MAE system and the FBC-CNN-MAE system are designed.Meanwhile,this thesis also proposes partial feedback scheme is designed to reduce the system transmit power and the ineffective loss without affecting the reliability of the signal.