Eigenstructure approach for array processing and calibration with general phase and gain perturbations

An eigenstructure approach for array processing with imperfect sensors is presented. Each sensor is modeled as having an unknown angularly independent gain error with an unknown angularly dependent phase error. The technique is capable of compensating for many sources of error: sensor location error, channel mismatch, unknown sensor phase patterns, and types of environmental inhomogeneities. The algorithm first determines estimates of the perturbed steering vectors, and then uses these to obtain angle of arrival (AOA) estimates. It can be used in calibration applications since the sensor perturbations are estimated in addition to the AOA of the sources. Each iteration of the algorithm consists of a Newton minimization step, but first a constraint is enforced by a mapping into the feasible set. Computer simulations are presented to illustrate the performance of the algorithm