Research And Experiment On Channel Estimation And Interference Cancellation Of High-speed Underwater Acoustic Communication System

As a huge treasure house of resources on the earth,the ocean has attracted the attention and research of various countries in recent years.However,because the underwater communication environment has the characteristics of multipath interference,fading channel,fast time-varying,strong noise,etc.,there are many difficulties in constructing a reliable and effective high-speed underwater acoustic communication system.Especially the presence of underwater channel response and impulsive interference seriously affects communication transmission.In order to ensure the high-speed transmission of signals,this thesis studies the channel response and impulsive interference of underwater channels.By accurately estimating channel response and impulsive interference,the communication quality of high-speed underwater acoustic communication systems can be greatly improved.Considering that Orthogonal Frequency Division Multiplexing(OFDM)has excellent characteristics against multipath propagation and inter-symbol interference,this thesis uses OFDM to build an underwater communication system model.On the basis of this model,this thesis conducts research on the sparse characteristics of underwater channel response and impulse interference.The main innovative content includes the following four points:First,by introducing the priori information of channel response and impulse interference,approximate message passing(AMP)algorithm can derive the posteriors of channel response and impulse interference.Then combined with the iterative process,the AMP-based channel estimation and interference cancellation algorithm proposed in this thesis can further improve the system performance.Second,because the smoothed l0 norm(SLO)algorithm uses approximate functions to approximate the l0 norm of the channel response and impulse interference,which makes the estimated values more accurate,this thesis designs a channel response and impulse interference estimation algorithm based on the SLO algorithm.Third,the super-resolution reconstruction convolutional neural network(SRCNN)can reconstruct a more accurate channel response by learning the high-precision channel response reconstruction process.Therefore,this thesis regards the channel response estimated by the SLO algorithm as a rough estimate,and designs the SLO-SRCNN model to further obtain an accurate estimate of the channel response.Fourth,considering that the real data used in this thesis is insufficient and the neural network has a large demand for data,this thesis uses the idea of transfer learning to design a channel estimation algorithm based on the SLO-SRCNN model.In this thesis,the proposed algorithms are evaluated through numerical simulations and real data.The real data was collected during a UA communication experiment conducted in the estuary of the Swan River,Western Australia.Through analysis and comparison of experimental results,it is proved that the proposed algorithms can better estimate the channel response and cancel the impulsive interference compared with the existing algorithms.The proposed algorithms can achieve lower bit error rate and frame error rate,thereby improving system performance.