The Research On The Acoustic Channel Identification Technology In Shallow Water Environment

The issue about underwater acoustic channel identification and equalization has been troubled many researchers in the acoustic field. The paper modeled for typical underwater acoustic channel and researched on algorithms of underwater acoustic channel blind identification and equalization with the above model. The main content and results are as follows:1. The research on blind identification and equalization has been analyzed, and so has the application status in the underwater acoustic field. Meanwhile, the key basics of blind identification and equalization have been introduced.2. The three adaptive algorithms, which are LMS, RLS and Improved Decision Feedback Blind Equalization algorithms, have been studied under the condition of underwater acoustic channel. The Improved Decision Feedback Blind Equalization algorithms are the improvement based on algorithms of Decision Feedback Blind Equalization and are the combination with Constant Modulus Algorithm (CMA), which results in improvement of performances. With the help of simulation and experiments on the sea, it comes to a conclusion that the three algorithms above, which are based on adaptive algorithms, couldn't identify channel under the condition of low signal transmission rate. It shows that narrowband signal can't be used to identify channel, especially with low signal-to-noise ratio (SNR). All the three algorithms above have effective distance of certain, and identify worse with a long distance. Compared to RLS and LMS, the Improved Decision Feedback Blind Equalization algorithms have the best identification results under the same conditions at the cost of large computation. While the LMS algorithm is the worst one among them.3. For underwater acoustic channel, three blind identification and equalization algorithms based on second-order statistics have been studied, which are OPDA, QR Decomposition, and Subspace Tracking Blind Equalization. After studying theory, improving algorithms and simulating, it comes to a conclusion that all the algorithms above have their own advantages and disadvantages and different appropriate occasions. The subspace algorithms need more stringent requirements on the conditions and the identification results are the best under certain conditions. The robustness of OPDA is better while its identification results are finite. The algorithms of QR Decomposition have the least dependence on the sample length than others, and have rapid convergence rate. However, the adaptation on other conditions and the identification results need to be improved. The performances of Subspace Tracking Blind Equalization are similar, for that they are the evolution of subspace algorithms, and they can identify channels under certain conditions. The results of convergence are satisfied after being equalized. What's more, the convergence rate is significantly improved, compared to subspace algorithms.4. The feature of multipath signal is studied by time-frequency analysis method under shallow water. Several linear and non-linear time-frequency analysis method is compared shallow water acoustic channel. RGK time-frequency analysis method is proposed to precess the istortional underwater acoustic signal. A kind of underwater acoustic channel identification model and the multi-source signal blind equalization is proposed. According typical sonar signal, deconvolution in the time-frequency field is carried out, and a kind of method of combining time-frequency field deconvolution with time field deconvolution is proposed. Each of method studied is verifid by simulation and trial data.