짧은 사이클을 가지지 않는 SC-LDPC 설계방법

Abstract

Spatially-coupled low-density parity-check codes (SC-LDPC) can be efficiently decoded under the sum-product algorithm. However, the decoding convergence may fail due to cycles in the Tanner graph. To improve the decoding convergence, we analyze the cycle properties of SC-LDPC codes, our analysis is based on the investigation of the connections among the monomials in the polynomial syndrome former of time-invariant SC-LDPC codes, which is the transposed parity-check matrix in polynomial form. Also, we propose a construction method of SC-LDPC codes without short cycles. We demonstrate that an increase of short cycles in the polynomial syndrome former may cause high bit-error-rates. Finally, we propose a novel method method to generate ``good" SC-LDPC codes with respect to their cycle properties