Direction of arrival estimation using robust complex Lasso

The Lasso (Least Absolute Shrinkage and Selection Operator) has been a popular technique for simultaneous linear regression estimation and variable selection. In this paper, we propose a new novel approach for robust Lasso that follows the spirit of M-estimation. We define $M$-Lasso estimates of regression and scale as solutions to generalized zero subgradient equations. Another unique feature of this paper is that we consider complex-valued measurements and regression parameters, which requires careful mathematical characterization of the problem. An explicit and efficient algorithm for computing the $M$-Lasso solution is proposed that has comparable computational complexity as state-of-the-art algorithm for computing the Lasso solution. Usefulness of the $M$-Lasso method is illustrated for direction-of-arrival (DoA) estimation with sensor arrays in a single snapshot case.