On the classification of radar returns using WVD, Welch and AR methods

A typical radar signal is non-stationary in nature in the sense that the spectrum of the signal comprises mainly clutter varying with time. These non-stationarities prevail even in the homogeneous clutter conditions. The classical spectral estimation methods have been applied for the classification of radar signals but were found to suffer from poor resolution. The application of stochastic modelling methods like AR and ARMA for the classification problem is found to be associated with inaccuracies such as showing ambiguous spectral widths, which is the most important factor to distinguish weather returns from birds, poor time resolution for time-varying clutter spectra, etc., despite being prominent in resolving spectral peaks. To provide an accurate description, the use of Wigner-Ville distribution (WVD) for the classification of radar returns, particularly for a groundbased radar, has been attempted. The paper proposes WVD, being a powerful tool for analysing the time-varying clutter like signals attributed by its excellent time resolution, as an alternative technique for the classification of radar returns. The analysis shows promising results, especially from the viewpoint of spectral widths, though faced with the problems of cross-spectral components. Examples of WVD applied to the simulated radar signal from an airport surveillance radar (ASR), are presented to illustrate the advantages of this method over the classical and stochastic modelling methods.