Turbo码迭代译码方法的改进及性能仿真分析

发布时间：2018-04-29 06:03

本文选题：Turbo码 + 性能仿真　；参考：《东北大学》2014年硕士论文

【摘要】：Turbo码是C.Berroi等人在1993年提出的一种新型的信道编码方案。它由于很好地应用了香农信道编码定理中的随机性编译码条件而获得了几乎接近香农理论极限的译码性能。Turbo码获得优异性能的根本原因是因为它采用了迭代译码的思想,但是采用迭代的方式进行译码,使得译码的计算复杂度相对于传统的卷积码来说要高,译码的时延也要大的多。改进和简化Turbo码的迭代译码,是当前Turbo码理论和实践研究的主要方向之一。本文在对Turbo码的编译码原理及算法进行了简单介绍的基础上,重点研究了Turbo码的改进迭代译码算法。在理论基础部分,首先本文对turbo码的基本原理进行了简单的介绍,接着对几种常用的译码算法进行了详细推导,最后提出两种改进的迭代译码算法。仿真部分本文利用MATLAB软件进行仿真分析,首先,本文对影响Turbo码性能的参数如分量码,归零处理方法,交织长度,码率,交织器,迭代次数,译码算法进行仿真,通过分析合理选择参数来提高Turbo码性能。基于对传统Turbo码译码算法的研究与分析,为了减少译码复杂度和译码延时,本文提出改进的迭代译码算法,即改进的组合译码算法和神经网络修正SOVA算法。改进组合译码算法是结合软输出维特比算法(Soft-Output Viterbi Algorithm, SOVA)和Log-MAP算法各自的优点,使得改进算法既能够达到Log-MAP算法的译码性能又能够减少译码复杂度。仿真结果表明所提出的改进迭代译码算法能够在降低译码计算复杂度的同时保持译码性能与传统的Log-MAP算法的性能近似。神经网络修正SOVA算法是对SOVA算法外信息计算的修正算法。传统的SOVA对判决的可信度的估计过高的问题,所以需要对软判决值△进行压缩,神经元网络的激励函数恰好符合具有连续平滑的压缩特性的修正函数。仿真结果表明SOVA算法获得更好的性能,而且在译码的复杂度方面与传统的SOVA算法相比差别不大,但是在译码性能上有很明显的提高,甚至接近Log-MAP算法的误码率。
[Abstract]:Turbo code is a new channel coding scheme proposed by C.Berroi et al in 1993. The fundamental reason for the excellent performance of Turbo codes is that it uses the random encoding and decoding conditions in Shannon's channel coding theorem well and obtains the decoding performance that is close to the Shannon's theoretical limit. The fundamental reason is that it adopts the idea of iterative decoding. But the iterative decoding method makes the computation complexity of decoding is higher than that of traditional convolutional codes, and the decoding delay is much larger. To improve and simplify the iterative decoding of Turbo codes is one of the main research directions in the theory and practice of Turbo codes. On the basis of a brief introduction of the encoding and decoding principles and algorithms of Turbo codes, this paper focuses on the improved iterative decoding algorithm for Turbo codes. In the part of theoretical foundation, the basic principle of turbo code is introduced briefly, then several common decoding algorithms are deduced in detail, and finally two improved iterative decoding algorithms are proposed. In the simulation part, we use MATLAB software to simulate and analyze the performance of Turbo codes. Firstly, we simulate the parameters that affect the performance of Turbo codes, such as component code, zeroing processing method, interleaving length, code rate, Interleaver, iteration times and decoding algorithm. The performance of Turbo codes is improved by analyzing the reasonable selection of parameters. Based on the research and analysis of the traditional Turbo decoding algorithm, in order to reduce the decoding complexity and decoding delay, this paper proposes an improved iterative decoding algorithm, that is, the improved combined decoding algorithm and the neural network modified SOVA algorithm. The improved combinatorial decoding algorithm combines the advantages of Soft-Output Viterbi Algorithm, SOVA) and Log-MAP algorithm, which can not only achieve the decoding performance of Log-MAP algorithm, but also reduce the decoding complexity. Simulation results show that the proposed improved iterative decoding algorithm can reduce the computational complexity of decoding while keeping the decoding performance similar to that of the traditional Log-MAP algorithm. The neural network modified SOVA algorithm is a modified algorithm for the computation of the information outside the SOVA algorithm. The traditional SOVA overestimates the reliability of the decision, so it is necessary to compress the soft decision value. The excitation function of the neural network coincides with the correction function which has the continuous smooth compression property. Simulation results show that the SOVA algorithm achieves better performance, and the complexity of decoding is not different from that of the traditional SOVA algorithm, but the decoding performance is obviously improved, even close to the BER of Log-MAP algorithm.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.22

【相似文献】