Phase/gain error compensation in sensor array via variational Bayesian inference

In this paper, direction-of-arrival (DOA) estimation with unknown gain/phase errors is considered in the context of sparse representation technique. This problem is formulated in a conjugate Bayesian model, where both the signal and phase/gain error are modeled statistically. All the parameters are estimated iteratively to achieve sparsity. The proposed approach requires no prior on the number of the sources. Moreover, the proposed approach can also provide superior performance in low SNRs and with large errors. The experimental results show that the proposed approach can outperform other reported methods in various scenarios.