Abstract:

To effectively acquire and sense signal which is sparse in the frequency domain, a super-resolution spectrum estimation method based on multirate co-prime sampling is proposed. Firstly, the signal is sampled via several sampling channels to construct the receiving model and the optimal sampling frequency selection criterion is given. Then the aliasing baseband of each sampling channel is estimated based on the super-resolution theory and the uniqueness theorem is derived. In the process of frequency spectrum recovery, an innovative support set reduction rule is proposed to further decrease the complexity of the candidate frequency support set. The proposed method can improve the estimation accuracy and resolution ability effectively as the reconstruction model is not influenced by the discretized grids, and the spectrum broadening effect arising from the truncated data is overcome. Simulation results demonstrate the correctness and effectiveness of the proposed method.