Sparse Bayesian learning approach for streaming signal recovery

We discuss the reconstruction of streaming signals from compressive measurements. We propose to use an algorithm based on sparse Bayesian learning to reconstruct the streaming signal over small shifting intervals. The proposed algorithm utilizes the previous estimates to improve the accuracy of the signal estimate and the speed of the recovery algorithm. Simulation results show that the proposed algorithm can achieve better signal-to-error ratios compared with the existing l1-homotopy based recovery algorithm.