摘要
通常用取最小绝对值方法对若干比特模二和的对数似然比(LLR)进行简化,该方法存在误差积累问题,因而不是最有效的.为此,提出了两种新的LLR简化算法:正比例函数拟和修正法和逐点平均值曲线修正法,并用这两种算法替代了低密度奇偶校验(LDPC)码归一化最有效可信传播(UMP-BP)译码中的LLR计算,使其在降低译码复杂度的情况下误码率更低.仿真结果表明,对于码长1 024 bits的LDPC码,采用所提出的LLR简化算法后性能较UMP-BP译码方法有0.4 dB提高,并与最优的可信传播算法接近,计算复杂度也有明显下降.
The method of taking minimum absolute value is generally used to simplify logarithm likelihood ratio （LLR） of modulo 2 sum of several bits. However, this method is not very effective because of that accumulates errors. Two novel simplified LLR calculating algorithms, the direct proportion approximation as well as the each dot approximation, were proposed and applied on uniformly most powerful -belief propagation （UMP-BP） algorithm, a decoding algorithm of low density parity check （LDPC） codes, to reduce the complexity and improve the BER performance. The simulation results show that, for an LDPC code of length 1 024 bits, the proposed method improves 0.4 dB in performance over the UMP-BP, and performs almost the same as the optimum belief propagation （BP） decoding algorithm.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2006年第1期46-49,共4页
Journal of Shanghai Jiaotong University
关键词
信道编码
低密度奇偶校验码
对数似然比
可信传播算法
迭代译码
channel coding
low density parity check （LDPC） codes
logarithm likelihood ratio （LLR） belief propagation （BP） decoding algorithm
iterative decoding
