AR spectral estimation with randomly missed observations

This paper represents a new spectral estimation method for time series with missed observations. An auto-regressive (AR) modeling approach is adopted. The AR parameters are estimated by optimizing a weighted mean-square error criterion. The method can be used in real-time adaptive contexts where the AR parameters are time varying. In general both regularly and randomly missed observations can be handled by this method. The spectral estimates are compared to those obtained by well known AR parameter estimators used in the cases where none of the signal samples is missed. The performance of the method is illustrated by some numerical examples.