Adaptive CFAR active sonar signal thresholding using radial basis functional neural networks

A recursive version of the adaptive constant false alarm rate (CFAR) sonar signal thresholding scheme using radial basis functional neural networks is proposed. Intensity thresholding has proven to be an effective technique to eliminate the low energy noise and to reduce the computational load in an underwater target tracking system. The proposed system has the following advantages: 1) the technique yields unbiased estimates under a nonhomogenous sea environment, because the false alarm rate is maintained at a constant level while the threshold changes with different sea environments; 2) the threshold for different range cells can be adaptively estimated since the noise under estimation is strictly local so that the received intensities of noise and targets are not affected by the distance the sonar signals travelled; and 3) the computational requirements are greatly reduced through the introduction of the recursive scheme