Research On Blind Separation Technique Of Target Echo And Reverberation

The existence of reverberation is one of the important problems which influence thedetection of underwater buried target. In order to isolate the target echo and reverberationfrom each other and to suppress the reverberation, it is expected to apply Blind SourceSeparation (BSS) into the study of anti-reverberation. So it is a significant work to study theblind separation technique of target echo and reverberation.Considering the signal model of active sonar, the target echo, reverberation and ambientnoise is analyzed and some basic assumptions are made. In this paper., a method based onspatial correlation is applied firstly to study the separability, which utilizes the difference ofthe spatial correlation radius of the target echo and reverberation. By computing thecorrelation coefficients of different sensors’ data, not only the statistical property is acquired,but also the input signal sources’ number is confirmed, which means the target echo andreverberation can be separated in some condition.Considering the properties of echo and reverberation, two blind separation methods (TFSand SONS) are applied into the real experimental data processing. The result shows that:Blind Source Separation algorithm can separate the target echo and reverberation in someextent. The separated echo has clearer time-frequency distribution and higher correlationcoefficients, which means BSS is not only feasible but also valid to separate target echo andreverberation.