An Optimized Algorithm to Construct QC-LDPC Matrix in Compressed Sensing

摘要

Aiming at the problems such as the large amount of data in transmission and difficulties in hardware implementation, an optimized algorithm is put forward to generate QC-LDPC measurement matrix based on limited geometry in compressed sensing, which can eliminate the short girth of 4 in Tanner graph through the design of basis matrix. Because of the quasicyclic characteristics, it can be realized by shift register so as to reduce the complexity of coding. The simulation results indicate that QC-LDPC matrix is superior to traditional measurement matrices by using the same OMP algorithm, and there are good improvements in the aspects of PSNR, SSIM, NMSE and runtime, which are conductive to the application of compressed sensing theory in real-time data transmission.