CAJ | 학술논문

针对喷泉编码的原始分组的度分布的统计，提出一种基于修正正态分布的编码算法。该方法提出两种统计模型，然后将编码简化为两个多重伯努利分布，发现当分布数目增大时，可以简化为正态分布。在这两种分布下，会有原始分组出现欠利用和过利用的情况。通过计算出这两个伯努利分布的期望值，并利用大量的仿真，统计其方差，得到度分布的置信区间。在编码过程中，能够自适应地将原先欠利用和过利用的部分进行调节，使得每个原始分组对于编解码的贡献趋于一致，提高纠错性能。实验表明，本文算法在高斯信道和删除信道下，与常规的喷泉编码相比较，误码率明显下降。
침대분천편마적원시분조적도분포적통계，제출일충기우수정정태분포적편마산법。해방법제출량충통계모형，연후장편마간화위량개다중백노리분포，발현당분포수목증대시，가이간화위정태분포。재저량충분포하，회유원시분조출현흠이용화과이용적정황。통과계산출저량개백노리분포적기망치，병이용대량적방진，통계기방차，득도도분포적치신구간。재편마과정중，능구자괄응지장원선흠이용화과이용적부분진행조절，사득매개원시분조대우편해마적공헌추우일치，제고규착성능。실험표명，본문산법재고사신도화산제신도하，여상규적분천편마상비교，오마솔명현하강。
An encoding algorithm based on the modified normal distribution is proposed for the statistics of the degree distribution in the original Fountain codes. Two mathematical modelsare put forward to simplify the codes to two multiple bernoulli distributions, and it's found that the two distributions can be simplified as normal distributions when the degree numbers increase. In the condition, the lack of use and the abuse of use will happen to the normal groups. The confidence intervals of the degree distributions are obtained when the expectationsof the two bernoulli distributions are calculated and the variances are counted by the extensive simulation. During the encoding process, the parts of the lack of use and the abuse of use can be adaptive adjusted that the contributions to the encoding and decodingprocedure tend to be uniform in the each of the originalgroups with an improvement in the decoding. The experimental results show that the proposed scheme has a significant decrease in the bit error rate in the Gaussian channel and the erasure channel compared with the present Fountain codes.