Abstract

This paper considers a two-dimensional direction-of-arrival (DOA) estimation problem from a collaborative, distributed antenna array where each subarray is a distributed sensing node that is arbitrarily oriented. While the relative locations of the subarrays are not precisely known, it is assumed that the configuration of each subarray is locally calibrated whereas the cross-covariance matrix between a pair of distributed nodes includes an unknown phase difference. Without explicitly estimating such unknown phase difference, subspace-based DOA estimation methods fail to coherently utilize the subarrays to locate the DOAs of the impinging signals. We propose a group sparsity-based approach to achieve accurate DOA estimation that is resilient to unknown phase disparities between subarrays. Simulation results clearly illustrate the effectiveness of the group sparsity-based approach using group LASSO, and the superiority over subspace-based methods, such as the MUSIC algorithm, is demonstrated.