Uncertainty and Impact Analysis of Probability Density Modeling for Ocean Reverberation
-
Abstract
Establishing an accurate reverberation probability density model is very important for signal detection in a reverberation background. First, the Gaussian distribution model, the Gaussian mixture distribution model, and the symmetric alpha-stable distribution model are used to model actual ocean reverberation data, respectively. The uncertainty in reverberation probability modeling is pointed out, and total deviation and the Kullback-Leibler divergence are used to measure this uncertainty. Then, taking the Gaussian mixture distribution model as an example and using the offset coefficient as an index, the impact of the uncertainty of reverberation modeling on the performance of the Locally Optimal Detector (LOD) is analyzed. Finally, a simulation analysis is carried out. Both theoretical and simulation results show that the uncertainty will lead to performance degradation of the LOD, and the stronger the uncertainty, the more severe the performance degradation.
-
-