Improved multiple signal classification algorithm for Direction of Arrival estimation based on covariance matrix of cross-correlation

In view of the poor performance of traditional Direction of Arrival (DOA) methods at low signal-to-noise ratios, an improved MUltiple SIgnal Classification (MUSIC) algorithm for DOA estimation applied to active detection system based on covariance matrix decomposition of cross-correlation (I-MUSIC) is proposed. Exploiting the transmission feature of active sonar, cross-correlation sequence between the transmitted signal and the array output is formulated. The spatial covariance matrix is then constructed from the sequence. Then matrix decomposition is implemented over the new spatial covariance matrix to estimate the DOA. It is proved that cross-correlation can suppress noise while preserving the phase information between array elements, which facilitate the subspace separation at low SNRs. Furthermore, another novel method based on correlation Time threshold (T-MUSIC) is proposed to further improve the DOA performance. Simulation results indicate that I-MUSIC and T-MUSIC can obtain a performance gain of 3 dB and 6 dB, with the estimate error being 77% and 53% of the original method respectively. Due to data selection via time threshold, T-MUSIC is not appreciably affected by noise, and thus outperforms IM-MUISC for 8 dB at low SNRs. I-MUSIC and T-MUSIC can improve the DOA performance at low SNRs significantly if applied to active multi-target detection system.