Abstract: An approximate l0 norm (AL0) based method is proposed for data compression and reconstruction. The method firstly uses sparse binary matrix as a measurement matrix, to compress and transmit the signal; At the received terminal, only the compression matrix and compressed signal are given, this paper uses the wavelet filter bank to design dictionary matrix, adopts the steepest gradient and projection method for searching the sparse signal representation. Finally, combining sparse signal representation and the designed dictionary matrix, this paper reconstructs the compressed data. At-sea experiments including chirp signal and single-carrier signal processing, NMSE, SNR and running time of several methods to evaluate the improved accuracy performance of the proposed method.