论文部分内容阅读
卷积码生成多项式对通信系统中信道编译码的性能有决定性影响,通常我们利用计算机搜索大约束长度的卷积码生成多项式,在搜索得到的生成多项式中有相当数量是恶性的需剔除,因此必须对搜索到的生成多项式进行检验。判定一个编码器的生成多项式是否是恶性的方法可以是数学推导或Matlab计算。但只适合约束长度较短的卷积码,而当约束长度较大时,运用现有的计算手段很难对编码器生成多项式进行检验。本文提出了一种新的基于GPU的高性能并行运算技术验证卷积码生成多项式的方法,GPU在进行大约束长度卷积码的恶码检验上有明显的速度优势。
Convolutional code generation polynomials have a decisive impact on the performance of channel coding and decoding in communication systems. Usually, we use computer to search for polynomials of convolutional codes with large constraint length. A large number of search polynomials are malignant to be eliminated. Therefore, The generator polynomials found must be verified. The method of determining whether an encoder’s generator polynomial is malignant can be mathematically derived or mathematically calculated. However, it is only suitable for constraining the convolutional codes with shorter length. However, when the constraint length is large, it is difficult to test the generator polynomial by using the existing calculation methods. This paper presents a new GPU-based high-performance parallel computing technology to verify the convolutional code generation polynomial method. GPU has a significant speed advantage in the verification of the bad code with large constraint length convolutional code.