In the sonar system, reverberation has always been the interference that is expected to be suppressed. For the underwater mobile sonar platform, the reverberation space-time spectrum may expand in the two-dimensional plane and has the property of space-time coupling, making trouble to the reverberation suppression.  

STAP (Space Time Adaptive Processing) is an effective method to suppress space-time coupling reverberation. However, the application of STAP in the underwater environment faces challenges caused by the complex underwater environment, insufficient samples of available uniform auxiliary data, and large amount of computation. 

In order to solve the above difficulties, researchers from the Institute of Acoustics (IOA) of the Chinese Academy of Sciences proposed a joint domain localized dimensionality reduced-dimension STAP method utilizing array symmetry characteristics. The method is able to reduce computational cost and improve the performance of reverberation suppression in the limited samples. 

The researchers studied the structure of reverberation covariance matrix, and discovered the properties of persymmetry. By combining it with JDL-STAP, they proposed a Per-JDL algorithm based on prior knowledge. 

The results showed that when there were enough training samples, the performance loss of JDL STAP compared with the optimal STAP was 6dB, while Per-JDL STAP compared with the optimal STAP was 4dB. When the training samples were insufficient, the performance loss of JDL STAP algorithm was 25dB, while that of Per-JDL STAP algorithm was 6dB. 

Fig. 1. Comparison of algorithm performance when the training samples were sufficient. (Image by IOA)  

Fig. 2. Comparison of algorithm performance when the training samples were insufficient. (Image by IOA)  

The algorithm maximized the use of training samples, reduced the computation complexity, and improved the performance of interference suppression. It has a good application prospect in the underwater mobile platform.  

The researchers will further verify the advantages of the method in practical application and improve the performance of the algorithm.