Sparse Reconstruction Based on Tanimoto Coefficient for DOA Estimation in Compressed Sensing

In traditional DOA (direction of arrival) estimation, the entire airspace in which the target signal is located needs to be searched, which greatly increases the computation, while traditional algorithms use internal product criteria to select atoms from redundant dictionaries that are not optimal and fail to ensure that the selected atomic energy is better suited to residual signals. This paper introduces compressed sensing theory, divides spatial domain into different regions based on transcendental knowledge, constructs heterogeneous overcomplete dictionary, and then reconstructs the original signal sparsely using atomic matching criterion based on Tanimoto coefficient. Two improved methods are proposed: (1) roughly determine the finite solution space with PM (Propagator Method) algorithm, then further define the initialization space according to CRLB (Cramer-Rao Lower Bound) and construct a nonuniform overcomplete dictionary based on the initialization space; and (2) propose a novel reconstruction of DOA signal based on Tanimoto coefficient similarity matching criteria. Simulation results show that at low SNR, the spatial resolution of reconstruction signal based on Tanimoto coefficient and nonuniform dictionary is higher than that of traditional uniform dictionary.