A Multiple-Measurement Vectors Reconstruction Method for Low SNR Scenarios

摘要

Multiple-measurement vectors (MMVs) compressed sensing (CS) provides a better reconstruction performance by using the jointly sparse property, but it suffers a severe performance degradation with the signal-to-noise ratio (SNR) decreasing. This brief presents, a novel MMV CS reconstruction method for the low SNR scenarios, called the signal combining-CS (SC-CS). Compared with other state-of-the-art reconstruction methods, the most innovative feature of the SC-CS is its ability of signal reconstruction in a weak SNR scenario. This makes it a promising candidate for many practical applications when the interfering noises are serious. The proposed method adopts the signal combining techniques, which can effectively suppress undesired noise in the combined measurement. In fact, the SC-CS provides a generalized MMV reconstruction framework in which different signal combining and reconstruction techniques can be used to augment this method. This framework also makes a good tradeoff between the reconstruction performance and the computational complexity. Simulation results show the effectiveness of the proposed method compared with the existing reconstruction methods especially for the low SNR scenarios.