A Modified IIN Algorithm for DOA Estimation Based on Sparse Representation

Recently, sparse representation has been widely used in localization and bearing estimation. The basic idea of the general sparse direction-of-arrival (DOA) estimation method is to divide the space into discrete grids. But the transmission signal's DOA doesn't always fall on the discrete network. This paper reformulates the above problem by exploiting the sparsity representation based on the modified iterative interpolation (IIN) algorithm. First, the linear combination of eigenvectors of the array covariance matrix is used in this paper. The multiple measurement vectors (MMV) can be converted to a single measurement vector (SMV) for sparse solution calculation in this way. And this method can reduce high computation of the MMV. Then, an on-grid DOA estimation can be got by the orthogonal matching pursuit (OMP) algorithm. In order to get an off-grid DOA which is closer to the real one, the modified IIN algorithm is simulated based on the on-grid DOA estimation which is obtained by the OMP algorithm. In the modified IIN algorithm, the most matched dictionary atom with the real signal DOA and the two neighboring atoms whose difference is semi gird resolution are chosen and the corresponding vectors are regarded as the measurement vectors. The smallest and sub-smallest Euclidean distances between the measurement vectors or its residual are applied to further improve the DOA estimation. The essential idea of the algorithm is a coarse-to-fine estimation. We demonstrate the effectiveness of the method on simulated data by comparing the estimator variance with the the L1-SVD algorithm. Simulation results show that our approach can reduce the DOA estimation error caused by grid effect and own low computation load.