LDPC 부호의 경판정 복호를 위한 전처리

Abstract

Low-density parity-check (LDPC) codes achieve the performance near to the Shannon limit by iterative decoding. They are extensively employed in communication system standards because of their capacity-approaching performance and low decoding complexity. The composition of this thesis can be divided into two parts. Firstly, we design LDPC codes that perform at rates close to the Shannon capacity. For this we summarize three techniques for analyzing LDPC code ensembles and calculate the noise threshold of regular LDPC codes in the BIAWGN channel. And then we optimize degree distributions using the differential evolution and construct the parity-check matrix using the progressive edgegrowth (PEG) algorithm. Secondly, we examine the structure of LDPC codes in the layered division multiplexing (LDM) system. And we propose the low-complexity decoding algorithm with the simple preprocessing which is applicable in some situations. We also propose the good puncturing criterion to generate rate-compatible LDPC codes. Simulation results show that the punctured codes using the proposed method have better performance compared to randomly punctured one.