Research On Positioning Error Correction In Underwater Acoustic Sensor Network With Sound Speed Inversion

The emergence of underwater acoustic sensing network(UASN)has greatly improved our potential to probe the marine environment.In most applications of UASN,node positioning is a fundamental and important task.Location information of nodes can help enhance the completion of other underwater task.The data will be useless if location information is missing in a location-related data collection tasks.Most of the existing underwater positioning models uses a single sound velocity value for distance estimation and node positioning,which neglects the influence of the stratification effect of water column and the dynamic variation of the sound speed characteris-tics.However,affected by the inhomogeneity of the underwater medium and the dynamic changes of the marine environment,there exists uncertainty in underwater positioning.Considering that,this thesis conducts research on sound speed estimation and positioning error correction in UASN.In the first part,the effect of each measurement noise[depth,Time of flight(ToF)and sound speed measurements]on the Cramer-Rao Bound(CRB)of range estimation is evaluated analyti-cally.It is shown that,similar to depth and ToF errors,sound speed uncertainty also plays a vital role in the CB.In addition,the dynamic nature of the marine environment can cause perturba-tion of sound speed profile(SSP),thus changing the path and propagation time of acoustic signal,which affects the stability of positioning in the end.It is necessary to provide more accurate SSP information for the positioning model,in order to reduce the impact of sound speed error on the performance of positioning in UASNThe second part of the thesis focuses on the development of accurate localization algorithms which can be applied in an underwater medium with a variable SSP.The analytical expression of the acoustic path between nodes is established through linear segmentation approximation of the known SSP.The sound path tracking between any two points is converted into a polynomial root finding problem.Finally,a time difference of arrival(TDoA)based underwater node position-ing optimization algorithm is proposed.Simulation and experimental data validation show that,compared to traditional methods,the proposed algorithm can achieve a more accurate and stable position estimation of underwater nodesThe third part of the thesis discusses the sound speed inversion and positioning error correc-tion in UASN.The linear relationship between signal propagation time perturbation and empirical orthogonal function(EOF)coefficients is established,based on the perturbation method and the EOF representation of the SSP.Simulation shows that,using the linear relationship,the average sound speed in the network coverage area can be solved by the inversion methods such as the least square method.By increasing the number of nodes and introducing a priori information for sequential filtering,the estimation performance can be effectively improved.Combined with the positioning model discussed in the second part,the distance estimation and TDoA positioning are taken as two examples to verify the performance improvement brought by sound speed estimation.Simulation results and the analysis of experimental data indicate that the target positioning error can be further reduced and better approach the CRB,with the update of inversion sound speed in the positioning model.In addition,the application of neural network methods in solving the inver-sion problem is also discussed in this part.Simulations show that in the case of solving sparse EOF coefficients,a neural network based model has more chance to obtain the global optimal solution.The model can obtain higher estimation accuracy while ensuring the resolution of the solutionThe research work of the thesis enriches theory and methodologies of the positioning in UAS-N.By using the signal propagation time disturbances between network nodes,the SSP can be inversed and updated for positioning models.Through the layered modeling of the sound ray path,a high-precision positioning algorithm for variable sound speed environment is established,which can give an unbiased estimate of underwater node position.Furthermore,the application of neural network model in solving sparse EOFs provides a new idea for improving traditional inversion algorithms.