Alternating projections-based gridless compressive beamforming with co-prime arrays

Abstract

Direction-of-arrivals (DOAs) estimation, beamforming, retrieves the angles of several sources from the outputs of receiving a sensor array. Compressive beamforming is a sparse signal recovery approach and has shown superior performance on DOA estimation. To overcome the angular searching grid issue, gridless techniques have been proposed. Most methods require a uniform linear array and use standard convex solvers that are computationally expensive. We propose a gridless compressive beamformer based on alternating projections. This method estimates DOAs by projecting a solution matrix alternatively. One projection works for measured-data-fitting, and the other works for having sparse DOAs. Our approach improves speed and accuracy and deals with arbitrary-shaped linear arrays. We validate the method using experimental data and test the DOA performance for a single snapshot, multiple snapshots with coherent arrivals, and co-prime arrays, a well-known non-uniform array.