Abstract: In order to improve the estimation accuracy of the existing DOA (Direction of Arrival) estimation
algorithms based on compressive sensing, a DOA estimation algorithm based on Improved Smoothed l 0 norm is
proposed. Firstly, we construct a proper smooth continuous function and give a proper decreasing sequence of
sets {δ} which is [δ 1 ,δ 2 ,…,δ J ] based on the initial solution of receiving data to get the minimum value of
F δ (S) for every single δ by using steepest descent method, secondly we take this δ as the original value for the
next iteration, and then we get the minimal solution of the smooth continuous function which is the imminent
minimum l 0 norm by a number of iterations. At last, we verify this proposed algorithm by simulation experiments.
The results denote that the proposed algorithm can achieve more excellent estimation performance than OMP
(Orthogonal Matching Pursuit)algorithm under the condition of single-snapshot and can get simple process and
high precision.

Key words: improved smoothed l 0 norm, steepest descent method,  array signal processing, direction of arrival (DOA) estimation