IN THE PRESENCE OF A CLASS OF UNKNOWN NOISE FIELDS
Surendra Prasad, Bindu Chandna · 1988
In the Signal Subspace algorithms for Direction-of-Arr ival estimation of signal wavefronts, the additive noise is assumed to be spatially white. For the case of completely unknown noise fields, Paulraj et a1 [2] and Prasad et a1 [31 have suggested covariance differencing techniques. When the noise is nonwhite but has a known covariance matrix, the problem could be handled through pre-whitening. Both these techniques have the disadvantage that they require an eigen-decomposition which is a computationally intensive technique. Also they require double the array aperture as compared to the situation when the noise is white. The purpose of this paper is to suggest a new approach for the use of signal subspace techniques to the case of unknown noise covariances, which is also computationally efficient.