A recovery algorithm of MWC sub-nyquist sampling based on random projection method

The existing recovery algorithm of modulated wideband converter (MWC)-based sub-Nyquist sampling is far from satisfactory. Aiming at this problem, a recovery algorithm for MWC based on random projection method is proposed. This algorithm projects the measurement value matrix of MWC onto a random matrix with lower dimension to form a new measurement value matrix, and then solves a multiple measurement vector problem using a solver proposed. The recovery performance is enhanced through examining and repeating the tentative solving processes. This paper validates the effectiveness of the algorithm from both theoretical and experimental perspectives. Numerical experiments demonstrate that, compared with the popular ReMBo algorithm, the proposed algorithm significantly improves the success rate of recovery. From the same number of channels a signal with more spectral bands can be recovered by this algorithm, and a signal with the same number of bands can be recovered using fewer channels. Furthermore, the run time of this algorithm does not increase greatly. In contrast, compared with ReMBo, this algorithm can use a lower time cost to achieve a higher recovery performance when the spectral bands of the signal exceed a specific number.