A GFK-Based SNR Adaptive DOA Estimation Method

This paper proposes a DOA estimation method based on geodesic flow kernel(GFK), which has signal-to-noise ratio(SNR) adaptability and good DOA estimation performance under the scene of low SNR in space communication. Different from other machine learning methods, the GFK-based DOA method does not need to use the training sample to train model in advance. In fact, this method uses transfer learning to treat the source domain sample and the target domain sample as two points in the high-dimension space, then looks for a geodesic manifold transform kernel function to transform the space of the source domain sample into the space of the target domain sample. After that we can use the direction-of-arrival label information of the source domain sample to estimate the direction of the test sample. The simulation results show that the proposed method has better accuracy under the same SNR than the traditional MUSIC method, meanwhile has low complexity.