Performance Evaluation of Several Adaptive Algorithms for Low and Zero-Doppler Active Sonar Signal Detection

摘要

The use of several stochastic gradient algorithms and of the deterministic least-squares lattice algorithm are investigated in this paper for the detection of low and zero-Doppler sonar signals buried in reverberation. Experimental results based on a large amount of real data are shown for different contact frequency shifts from the reverberation centre frequency with a fixed signal-to-reverberation ratio equal to 0 dB. Among the different algorithms, the least-squares lattice and the least-mean lower will be shown to outperform the others and zero-Doppler contact detection will be possible in many instances with the least-squares lattice performing best.