Applications Of Sparse Reconstruction Algorithms Of The Compressive Sensing Theory In Underwater Acoustic Communication And Array Signal Processing

Compressive sensing theory is a new theory of signal processing which has been developed in recent ten years.The sparse reconstruction algorithms of compressive sensing can be used not only for signal sparse reconstruction but also for signal parameter estimation in problems with sparse structure,which is of great value.In array signal processing problems,such as DOA estimation and source localization based on matched-field processing,as well as in channel estimation,the signal often has sparse characteristics,but the conventional parameter estimation methods can not give sparse estimation results.In this thesis,several greedy algorithms of compression sensing theory are studied,and their application in DOA estimation,source localization based on matched-field processing,and channel estimation are studied.The main contents are as follows.Firstly,the basic frame and mathematical model of the compressive sensing theory are introduced,and several sparse reconstruction algorithms based on the greedy idea,such as MP,OMP,OMP-DCD,and MMP,are introduced.This lay the foundation for the follow-up study about signal parameter estimation problems with sparse characteristics.Secondly,the classical DAS algorithm and Capon algorithm are introduced,and the application of sparse reconstruction algorithms in DOA estimation is studied.For the source localization based on matched-field processing,the Bartlett algorithm and the MFP-PDS algorithm are introduced.The OMP-DCD algorithm is adopted to solve the problem that the MFP-PDS algorithm can not locate multi-sources.The sparse double-spreading channel estimation technique based on the OMP algorithm and the space-time-frequency joint channel estimation technique based on the SA-OMP algorithm are introduced.Finally,the theoretical simulation and experimental data processing are conducted.The performance of several DOA estimation algorithms,source localization algorithms based on matched-field processing,sparse double-spreading channel estimation algorithm and the joint channel estimation algorithm for the time delay,the DOA,and the frequency are investigated by simulation.The results of SWellEx-96 sea experiment show that the proposed OMP-DCD-PDS algorithm has strong ability in multi-source localization based on matched-field processing.The results of the South China Sea in December,2015 show that,according to the DOA estimation results,the symbol error of the system is reduced from12.82% to 0.08% compared with the scheme without spatial filtering.That is,the combinationof DOA estimation and spatial filtering to deal with single-carrier communication data in sea trials can greatly improve the performance of communication systems and reduce the bit error rate.