Likelihood ratio detection of signals on reverberation noise

In the context of stochastic dependency of the noise and signal - as in the case of the reverberation noise - the classical methods of signal detection are not applicable. The paper presents a likelihood ratio detection method based on stochastic calculus. The novel aspects of the method are: the signal and the noise need not be stochastically independent and they both can be nonstationary and non-Gaussian. This method is applicable to the active sonar which is a reverberation limited system. Simulations related to the underwater propagation of the acoustic wave show a good performance of the detector at the receiver operating characteristic level. The likelihood ratio detection method can also be successfully applied in mobile communication for the channel dynamical allocation, as well as in radar applications.