Parameter estimation for McDaniel´s non-Rayleigh reverberation model

Reverberation in active sonar systems has often been observed to follow non-Rayleigh distributions. Advanced signal processing algorithms can account for such a deviation by assuming a particular distribution and estimating its parameter values from observed data. McDaniel´s (1990) model has shown promise as being capable of accurately representing real data with a minimal parameterization. This paper first presents methods for approximating the cumulative distribution function of McDaniel´s model and then a comparison between method of moments (MME) and maximum likelihood (MLE) parameter estimation. It was observed that the MME and MLE provide similar performance. The MME has the advantage of a simple and rapid implementation and the disadvantage of a non-zero probability of a solution not existing. The MLE will always provide a solution; however, it requires multivariate optimization. The frequency with which the MME solution does not exist increases as the amount of data used in the estimation decreases. In such cases where estimation with minimal data is required, it is shown how techniques such as bootstrapping or pruning can provide a solution