New results on linear prediction for clutter cancellation

Some moving-target indicator (MTI) techniques are reviewed and those based on the linear prediction theory are examined. A description is given of the effects of the position of the reference sample in an MTI processor on cancelling clutter in a radar system. The analysis of this generalized linear prediction MTI is based on the evaluation of the receiving operating characteristics (ROCs), assuming that the clutter covariance matrix is known a priori as well as the matrix being estimated on line. It is shown by analytical evaluations and Monte Carlo simulations, that the performance is generally insensitive to the position of the reference sample. An additional analysis is carried out to compare the performance of the generalized linear prediction MTI with the MTI based on the Hsiao approach. It is seen that with the formation of the interference of two superimposed clutter sources the generalized linear prediction MTI suffers, in some cases, several decibels of loss.<>