Direction-finding with sensor gain, phase and location uncertainty

The estimation of the directions-of-arrival (DOAs) of multiple narrowband sources impinging upon an arbitrary array of sensors is discussed. Emphasis is directed to applications in which the sensor locations are only nominally known and the sensor gains and phases may be in error. The signal subspace method combined with a QR-based Gauss-Newton algorithm is used to estimate the DOAs, as well as the sensor gain, phase, and location of each sensor simultaneously. This approach is shown to outperform conventional high-resolution DOA algorithms. In addition, two Cramer-Rao bounds are derived which treat the cases where the sensor gains and phases are considered as random parameters with known density distributions, and the sensor locations are considered as unknown parameters