Off-Grid Parameter Estimation Based on Joint Sparse Regularization

Abstract:
We present a new off-grid parameter estimation method based on joint sparse regularization. More specifically, we consider the application of Direction-Of-Arrival (DOA) estimation from Multiple Measurement Vectors (MMVs). The MMV based DOA estimation problem exhibits joint sparsity which is commonly exploited by mixed-norm regularization, e.g., the l2,1 mixed-norm, which is known to suffer from high computational complexity in case of many MMVs or large dictionary matrices. To overcome the computational complexity associated with mixed-norm minimization the SPARse ROW-norm reconstruction (SPARROW) approach has recently been introduced as an equivalent, but less complex problem formulation. For the special case of Uniform Linear Arrays (ULAs) or ULAs with missing sensors, the GridLess-SPARROW (GL-SPARROW) has been presented. In this paper, we extend the SPARROW formulation by an off-grid estimation method for arbitrary array topologies, based on linear interpolation in form of first order Taylor expansion.