ADAPTIVE ARRAY PROCESSING USING PREDICTED COEFFICIENTS AS CONSTRAINED CONDITIONS

The algorithm for linearly constrained adaptive processing has outstanding advantages, but in this algorithm, the frequency characteristics o?the array signal in the direction of interest must be given and the constrained conditions be derived from the given frequency characterictics.
When the frequency characteristics is unknown, a new method for adaptive array processing is described. In the practical situations, it is easy to obtain the estimation of the power spectrum in the observed direction instead of the frequency characteristics. Then the maximum entropy spectral analysis can be incorporated with the adaptive array processing. The autocorrelation function can be calculated from the power spectrum and the predicted coefficients of MEM can be obtained by solving the normal equation of the correlation matrix. The predicted coefficients can be used as the constrained conditions of the algoithm for the adaptive array processing..
In this paper a new method of array processing is presented for the more general cases, in which the signal characteristics in the observed direction are completely unknown. Firstly, the space filtering of the adaptive array is performed by the all-pass constrained conditions in frequency domain and the power spectrum in the output of the array processor is estimated. Secondly, by using the above mentioned method, the space-frequency domain optimum filtering is performed by the adaptive array processor using predicted coefficients as constrained conditions.
A series of computer simulation experiments have been conducted, and the results are presented at the end of the paper.