Neural structure for direction finding with sensor array uncertainties

This paper presents an auto-calibration technique to estimate the DOAs of unknown signals in the presence of sensor gain and phase errors. Compared with other auto-calibration algorithms, this algorithm is computationally more efficient, and has a unique solution for a linear array with the imposed constraints. Furthermore, the complex computation in the objective likelihood gradients is performed by neural networks. An analogue scheme encompassing neural networks has been proposed for the algorithm's implementation.