基于删除信道下LT码的概率密度研究  

Research on Probability Density of LT Code Based on Delete Channel

作　　者：张娅 牛芳琳[1] 金晶晶 Zhang Ya;Niu Fanglin;Jin Jingjing(School of Electronics and Information Engineering,Liaoning University of Technology,Jinzhou,121000,China)

机构地区：[1]辽宁工业大学电子与信息工程学院,辽宁锦州121000

出　　处：《信息通信》2020年第12期44-49,共6页Information & Communications

基　　金：辽宁省自然科学基金指导计划项目(201602373)。

摘　　要：由于噪声的随机性与度1符号产生与分布的随机性,且必须成功接收足够数量的独立编码符号才能恢复源符号,所以LT码在删除信道中译码符号数量具有不确定性,那么其概率密度分布也随着信道概率的变化而变化,针对LT码在删除信道下译码符号概率密度分布函数的问题,提出了一种基于高斯混合模型与K-均值结合的改进聚类算法。该算法首先利用K-均值完成数据初始划分,并根据分类后的聚类参数给出EM算法迭代的初始值;然后利用EM算法学习GMM,完成模型建立。实验结果表明:通过建立GMM,可以得到LT码在删除信道下概率密度分布函数。Due to the randomness of noise and the randomness of generation and distribution of degree 1, and a sufficient number of independent coded symbols must be successfully received to recover the source symbol, the number of decoded symbols in the LT code is uncertain in the deletion channel, so its probability density distribution also changes with the change of channel probability.Aiming at the problem of the decoding symbol probability density distribution of the LT code in the deleted channel, this paper proposes an improved clustering algorithm based on the combination of Gaussian mixture model and K-means. The algorithm first uses Kmeans value completes the initial data division, and gives the initial value of the EM algorithm iteration according to the classified clustering parameters;then uses the EM algorithm to learn GMM and completes the model establishment. The experimental results show that by establishing GMM, you can get the LT code probability density distribution function in the delete channel.

关 键 词：LT码 概率密度公式 K均值算法 高斯混合模型 EM算法 

分 类 号：TN911.22[电子电信—通信与信息系统]