A Novel Soft-Output Demapper for DOCSIS 3.1

Abstract

DOCSIS 3.1, the new international standard for high-speed cable television (CATV), uses Low Density Parity Check (LDPC) code for forward error correction (FEC). It is widely known that accurate soft information is critical to the performance of LDPC decoding, but existing bit log-likelihood ratio (LLR) expression only works well under high code rate. In this paper, an improved bit LLR expression is proposed to compensate the difference between the existing LLR expression under low code rate for Gray coded M-ary quadrature amplitude modulation (QAM). In addition to high accuracy, the improved LLR is simple and efficient to manipulate. Besides LDPC, the improved LLR is also applicable to other codes that require soft information. Simulation results show that the improved LLR expression achieves a significant improvement in performance for both LDPC and Turbo decoding.