A sparse recovery approach for direction finding in partly calibrated arrayscomposed of subarrays with unknown displacements is introduced. The proposedmethod is based on mixed nuclear norm and 1 norm minimization and exploitsblock-sparsity and low-rank structure in the signal model. For efficientimplementation a compact equivalent problem reformulation is presented. The newtechnique is applicable to subarrays of arbitrary topologies and grid-basedsampling of the subarray manifolds. In the special case of subarrays with acommon baseline our new technique admits extension to a gridlessimplementation. As shown by simulations, our new block- and rank-sparsedirection finding technique for partly calibrated arrays outperforms the stateof the art method RARE in difficult scenarios of low sample numbers, lowsignal-to-noise ratio or correlated signals.
展开▼
