Parameter Estimation for Real Filtered Sinusoids

摘要

This research develops theoretical methods for parameter estimation of filtered,pulsed sinusoids in noise and demonstrates their effectiveness for Electronic Warfare (EW) applications. Within the context of stochastic modeling, a new linear model, parameterized by a set of Linear Prediction (LP) coefficients, is derived for estimating the frequencies of filtered sinusoids. This model is an improvement over previous modeling techniques since the effects of the filter and the coefficients upon the noise statistics are properly accounted for during model development. From this linear model, a relationship between LP coefficient estimation and Maximum Likelihood (ML) frequency estimation is derived and several coefficient estimators, based on fixed point theory and ML techniques, are constructed. A bound for the coefficient estimation error is developed and used to gauge the quality of point estimates directly from the data and knowledge of the noise variance. Furthermore, a multirate implementation of an EW digital channelized receiver is described functionally and probabilistically. When applied to the EW receiver, simulations indicate the new estimators provide unbiased, minimum variance, parameter estimates of filtered sinusoids at lower SNRs than the estimators currently employed. The bounds on the estimation error are then used establish confidence intervals for each point estimate.