CAJ | 학술논문

基于随机计算实现了二进制低密度奇偶校验码(LDPC)译码初始化的公式.首先对 LDPC译码中的初始化公式进行坐标映射,将输入数据转换到概率域上,然后对其进行线性近似,最后采用随机计算和门电路实现初始化,以达到降低硬件实现复杂度的目的.该方法与已有的存储器方法相比硅片面积减少了37%,并且在保持译码性能基本相同的情况下译码硅片面积最大减少了12.3%,当使用 FPGA 实现时,利用该方法使得相应的4输入查找表个数减少了12.8%.
기우수궤계산실현료이진제저밀도기우교험마(LDPC)역마초시화적공식.수선대 LDPC역마중적초시화공식진행좌표영사,장수입수거전환도개솔역상,연후대기진행선성근사,최후채용수궤계산화문전로실현초시화,이체도강저경건실현복잡도적목적.해방법여이유적존저기방법상비규편면적감소료37%,병차재보지역마성능기본상동적정황하역마규편면적최대감소료12.3%,당사용 FPGA 실현시,이용해방법사득상응적4수입사조표개수감소료12.8%.
A low-complexity initialization implementation in stochastic LDPC (Low-density-parity-check) decoding is proposed. The initialization formula in the LDPC decoder is transformed by piecewise linear approximation, and the input data from the channel is mapped into the probability domain and then processed by the stochastic computation. As compared with the existing lookup table method, the silicon area of the proposed method is reduced by 37%, and the overall area of the decoder is reduced by 12.3%. The proposed method could reduce the number of 4-input lookup table by 12.8% in FPGA implementation.