SPARSE SIGNAL RECONSTRUCTION FROM LIMITED DATA USING BAYESIAN COMPRESSIVE SENSING VIA BELIEF PROPAGATION

Prateek Paliwal, Manish Sharma
2019 International journal of electronics engineering and application  
Compressive sensing (CS) is a novel paradigm for acquiring signals, based on the idea that one can efficiently capture all the information of sparse domain by sampling only a part of signal through sub-Nyquist signal acquisition. As Bayesian inference is substituting conventional CS methods, the work employ Bayesian inference which depict CS matrix by a factor graph to accelerate both encoding and belief propagation (BP) decoding. Two state mixture Gaussian model is used to model prior for
more » ... odel prior for sparse signal. To decode a length-N signal having K large coefficients, our BCS-BP decoding algorithm uses O (K log (N)) and O (N log2 (N) ) computation. BCS-BP algorithm is easily versatile to various signal models.
doi:10.30696/ijeea.vii.ii.2019.24-36 fatcat:khpxcigo4rfzrnxxq3nl44z3ui