Wideband Direction of Arrival Estimation of Optimal Covariance Based on Modified Squared TOPS

Abstract

In the actual communication scenario, traditional algorithm using subspace orthogonality has poor performance in wideband Direction of arrival(DOA) estimation. Especially after being disturbed by non-uniform noise, low SNR, and small snapshots in the real environment, the orthogonality of subspace is difficult to be satisfied, and the performance of DOA estimation often deteriorates sharply. A new wideband DOA estimation method of optimal covariance based on modified squared TOPS is proposed. The new method uses the modified matrix decomposition algorithm to solve the coherence and reconstructs the noiseless covariance matrix with convex optimization. Then it is combined with TOPS algorithm. The signal subspace projection is introduced to replace the null space projection, and the decision matrix is modified by using the square TOPS method. Finally, the DOA estimation is completed by finding the trace of the decision matrix to replace the judgment of the lack of rank. The simulation results show that, the new method has lower side-lobes and sharper discernible spectral peaks. It improves the estimation accuracy of wideband signal effectively, reduces the interference from non-uniform noise, and shows better resolution and accuracy.