Direction-of-arrival estimation using a mixed l2,0norm approximation

  • Authors:

  • Md Mashud Hyder;Kaushik Mahata

  • Affiliations:

  • Department of Electrical Engineering, University of Newcastle, Callaghan, NSW, Australia;Department of Electrical Engineering, University of Newcastle, Callaghan, NSW, Australia

  • Venue:
  • IEEE Transactions on Signal Processing
  • Year:
  • 2010

Quantified Score

Hi-index 35.68

Visualization

Abstract

A set of vectors is called jointly sparse when its elements share a common sparsity pattern. We demonstrate how the direction-of-arrival (DOA) estimation problem can be cast as the problem of recovering a joint-sparse representation. We consider both narrowband and broadband scenarios. We propose to minimize a mixed l2,0 norm approximation to deal with the joint-sparse recovery problem. Our algorithm can resolve closely spaced and highly correlated sources using a small number of noisy snapshots. Furthermore, the number of sources need not be known a priori. In addition, our algorithm can handle more sources than other state-of-the-art algorithms. For the broadband DOA estimation problem, our algorithm allows relaxing the half-wavelength spacing restriction, which leads to a significant improvement in the resolution limit.