Statistical analysis of phase parameter estimates of random amplitude linear FM signals

In radar and sonar, it is possible to model the return signal as a random amplitude, frequency-modulated signal in additive noise. The use of cyclic moments for estimating the phase parameters of random amplitude linear FM signals in additive noise is investigated. In particular the covariance matrix of the phase parameter estimates obtained from the cyclic moments is derived. It is shown that, in general, the accuracy of the phase parameter estimates is dependent on the lag parameter chosen for the cyclic correlation. The analysis also reveals that, under certain conditions, the variances of the lower order phase parameter estimates are independent of the lag parameter.