An Improved Depth-First Multipath Matching Pursuit Algorithm Based on a Multi-Atom Selection Strategy

The depth-first multipath matching pursuit (MMP-DF) algorithm has received much attention for sparse signal reconstruction in compressed sensing. In this paper, we propose an algorithm referred to as depth-first multipath generalized matching pursuit (MGMP-DF) that investigates a multi-atom selection strategy to update the candidates. Multiple atoms selected per iteration decreases the number of iterations and improves the chance of selecting the true support. Simulation results show that the proposed algorithm outperforms some state-of-the-art works in terms of reconstruction performance.