A beamspace sparse-overcomplete signal representation method for bearing estimation by sensor array

Recently, the sparse overcomplete signal representations have been introduced to solve the direction-of-arrival (DOA) estimation problem by sensor array. However, it still has high computational complexity, and as the dictionary, i.e. a collection of possible direction response vectors, is required to be ldquoovercompleterdquo, it has to cover all the possible DOAs both of the interesting signals and uninteresting interferences to get the signal DOA estimation. This not only takes more program waiting time, but also makes the results unstable when the arrival angles of interferences are varying rapidly. In this paper, with the extension of the sparse overcomplete representation method (M-SBL) due to D. P. Wipf and B. D. Rao in beamspace bearing estimation, the beamspace M-SBL (BM-SBL) method is proposed. Since its dictionary contains only the possible DOAs of interest and taking advantages of beamspace conversion, the proposed method takes much less computational cost. On the other hand, simulation results show that the BM-SBL outperforms its element space counterpart M-SBL in both resolution and accuracy.