中短长度LT码的展开图构造方法

Unfolding Graphs for Constructing of Short and Moderate-Length LT Codes

该文提出了中短长度LT码的展开图构造方法。该方法以输出节点为根展开Tanner图,当满足停止准则后,随机选择未被覆盖或最深展开层的输入节点与根相连。展开图法通过展开深度门槛和最大可展深度两个参数的设置,在控制短环的基础上,在尽可能大的范围内随机地选择输入节点。仿真结果表明,该方法与随机图等方法相比,能够有效地改善中短长度LT码的性能。

This paper proposes one construction method for short and moderate-length Luby Transform （LT） codes by unfolding Tanner graphs from every output node as root. For each output node, when the stopping criterion is satisfied, one uncovered input node or one input node on the deepest layer is randomly selected and connected to the root. Two parameters, the unfolding depth threshold and the maximal unfolding depth, are adjusted to control the girth of short cycles and make the range of random selection as large as possible. Simulation shows the performance improvements of the proposed algorithms are obvious compared with the random graph method for short and moderate length LT code.