exploiting this information.
When the array is calibrated precisely, the natural We consider the problem of estimating the steering vecway of estimating the steering vectors is to first estitors of an uncalibrated diversely polarized array. The mate the directions of arrival and then form a linear array elements are assumed to consist of several groups. combiner from the corresponding steering vectors to In each group the elements have the same polarization sensitivity and the same unknown gain pattern, up to an separate the signals. In practice, it is difficult to unknown multiplicative factor. The phases of the elements maintain a precisely calibrated array. Temperature, are arbitrary and unknown. We identify a cost function pressure, humidity, mechanical vibrations, and obwhose minimizer is a statistically consistent and efficient jects in the near field all affect the calibration preciestimate of the unknown parameters. An iterative algosion. rithm for finding the minimum of that cost function is It is known that ''blind'' estimation is possible for presented. The proposed algorithm is guaranteed to connon-Gaussian signals by using high-order statistics verge. The estimated steering vectors are used for conof the received data [7-9]. The term ''blind'' is used structing a matrix whose multiplication with the observed to indicate that the array manifold is assumed to be data yields an estimate of each of the signals. A comparicompletely unknown. Techniques based on secondson of Monte Carlo experiments and theoretical bounds is order statistics cannot solve the ''blind'' estimation used to validate the analysis.