CAJ | 학술논문

LDPC码分层译码算法在进行整数量化操作时，存储单元的限制会导致译码信息的溢出。本文系统分析了溢出错误的原因，并提出了两种改进的分层译码方案，错误部分消除方案（ partially eliminating errors scheme，PEES）和不同比特量化方案（different bit quantization scheme，DBQS）。两种改进方案分别从消除部分错误和避免溢出错误的角度来改进译码性能，且硬件实现时只需增加一定数量的加法器和移位操作。通过对不同码长、不同量化比特的LDPC码进行仿真，结果表明，2种方案均有效地抑制了溢出错误，与基于全精度浮点数运算的修正算法相比，在误码率为1．0×10－4时，分别仅有约1 dB和1．8 dB的性能损耗。
LDPC마분층역마산법재진행정수양화조작시，존저단원적한제회도치역마신식적일출。본문계통분석료일출착오적원인，병제출료량충개진적분층역마방안，착오부분소제방안（ partially eliminating errors scheme，PEES）화불동비특양화방안（different bit quantization scheme，DBQS）。량충개진방안분별종소제부분착오화피면일출착오적각도래개진역마성능，차경건실현시지수증가일정수량적가법기화이위조작。통과대불동마장、불동양화비특적LDPC마진행방진，결과표명，2충방안균유효지억제료일출착오，여기우전정도부점수운산적수정산법상비，재오마솔위1．0×10－4시，분별부유약1 dB화1．8 dB적성능손모。
When layered decoding algorithm for LDPC (Low-Density Parity-Check) codes is performing integer quantization operation, restriction of storage units will cause the problem of decoding information overflowing.This paper systematically analyzes the reason why overflow errors occur, and proposes two improved layered decoding schemes:Partially Eliminating Errors Scheme (PEES) and Different Bit Quantization Scheme (DBQS).Two im-proved schemes improve decoding performance by partially eliminating some errors and avoiding overflow errors re-spectively, and hardware implementation requires only some extra summators and shift operations.In this paper, we simulate the performance of the LDPC codes in different code lengths and different quantization bits, the simulation results and their analysis show preliminarily that:both programs both effectively suppress overflow errors;compared with the modified algorithm based on full-precision floating-point operations, the two proposed schemes can achieve a performance loss of about only 1 dB and 1.8 dB respectively around BER=10-4 .