Research On Doa Estimation With Short Snapshots

In modern military/satellite communications, the widely used of frequency hopping signals or short burst signals from complex electromagnetic environment, impacts that the array antenna can not get enough snapshots. Subspace-based high resolution DOA algorithms, require a correlation matrix which is always dependent on the statistics of array measurement. Therefore, these algorithms are influenced by the well-know threshold effect. It means that the performance of these algorithms deteriorate when the number of snapshots or the signal-to-noise ratio(SNR) drops below a certain threshold. When the number of snapshots is limited, the statistics is not quite accurate, which directly results in the overlap between the signal subspace and the noise subspace. In this paper, the effect of limited snapshots on the performance of the traditional subspace-based algorithms are analyzed. The DOA algorithms with limited snapshots which are the main contents of this paper. The creative works are concluded as follows:The suggested algotirhm is based on the spatial smoothing with improved aperture. Through SSIA, the algotirhm expolits the shift invariance property of the array steering matrix and reconstructs the data matrix without loss of array aperture. The eigen-subspace is obtained by new data matrix without requiring EVD and the priori knowledge of source number. It shows by Monte-Carlo simulations that the SSIA Power ESPRIT has the same performance with SSIA ESPRIT. The most outstanding advantage of SSIA Power ESPRIT is that it can resolve the DOAs from single snapshotswithout knowing the number of sources accurately. Under non-cooperative communications, the proposed algorithm can be an important technique.The proposed algorithm, PR UQR-OPM, is suitable for small snapshots by pseudo-noise resampling technique. The pseudo-noise resampling technique makes it possible to improve the accuracy of DOA estimation. It use some local performance test after each resampling run to exclude the outlier-estimators from the whole resampled ones. In order to improve the methods without EVD, QR-OPM is choosen as the new propagator. Taking advantage of the unitary Root-MUSIC to calculate the DOA estimation translate complex arithmetic into real one. Therefore, the proposed algorithm has lower computational complexity, and the performance of the algorithm is same as PR Root-MUSIC. This makes the new algorithm well applicable for the situations which require DOA estimation of fast moving sources.