Considering near-field sources localization

a method based on sparse signal reconstruction is presented. The sparse optimization problem is solved through minimizing L1-norm of sparse signals

and the sources localization is realized. The method uses an estimate of noise power as the tradeoff of the optimization problem

without prior knowledge on noise. To reduce the computational complexity

the two-dimensional positioning problem is transformed into two one-dimensional parameter estimation problems. Simulation results show the performances of the proposed method.