Turbo码交织器与迭代译码停止准则研究

发布时间：2018-10-18 21:32

【摘要】：Turbo码是一种新的信道编码技术,它的出现为信道编码理论和实践带来了一场革命,特别是在抗衰落、抗干扰能力方面有着传统信道编码技术无法比拟的优势,因此在INMARSAT-phone M4系统、CDMA多用户检测和第三代移动通信系统IMT-2000中得到了广泛的应用。但Turbo码也存在着一些亟待解决的问题,需要人们进一步的研究解决。本文对Turbo码的交织器和迭代译码停止准则进行了积极地研究,具体内容如下：首先,对Turbo码编码和译码的基本原理进行了系统分析,同时介绍了LOG-MAP和SOVA两种迭代译码算法。在此基础上,利用MATHLAB软件针对影响Turbo码性能的参数(即分量码、交织器、删余器、迭代次数、译码算法)进行仿真,通过分析仿真结果给出了提高Turbo码性能的合理参数。其次,对Turbo码交织器进行了研究,并针对螺旋交织器中的不足提出了一种新型半规则交织器。该新型交织器采用随机数组将交织矩阵的行列顺序由固定变为随机,同时采用正螺旋顺序读出的方式,使交织序列具有模2特性,既增大了相邻比特交织后的距离,又解决了经过删余矩阵后造成的对信息位保护不均匀的问题。为了使半规则交织器达到最佳工作状态,即把行列重置数组作为待解集,利用遗传算法进行搜索,从而找到最佳的行列重置数组。仿真结果表明,无论是在长帧还是短帧情况下,半规则交织器的性能都要优于分组交织器、螺旋交织器和随机交织器,与S型随机交织器相接近。经过优化,半规则交织器的性能更是保持在了最优状态,在相同信噪比条件下的误码率也获得了很大增益。最后,为了减小迭代译码延时、降低平均迭代译码次数,研究了Turbo码迭代译码停止准则,并提出了两种改进的迭代译码停止准则,即改进的HDA准则和改进的SDR准则。改进的HDA准则定义了一个基于外部信息的不可靠度量,当这个不可靠度量达到门限值时,判定本帧译码不能收敛,即可提前停止译码,从而解决了在低信噪比条件下HDA准则失效的问题；改进的SDR准则则采用连续两次迭代输出的符号差别数比值来代替原来的判决函数,从而解决了原始准则在遇到不收敛帧时不能停止译码的问题。仿真结果表明,在几乎不降低译码性能的情况下,两种改进准则的平均迭代次数都比原始准则有了明显的减少。
[Abstract]:Turbo code is a new channel coding technology, which brings a revolution for channel coding theory and practice, especially in anti-fading, anti-interference ability has the advantage that the traditional channel coding technology can not compare. Therefore, it has been widely used in INMARSAT-phone M4 system, CDMA multi-user detection and third generation mobile communication system IMT-2000. However, there are still some problems to be solved in Turbo codes, which need to be further studied and solved. In this paper, the Interleaver and the stopping criterion of iterative decoding for Turbo codes are studied actively. The main contents are as follows: firstly, the basic principles of coding and decoding of Turbo codes are systematically analyzed, and two iterative decoding algorithms, LOG-MAP and SOVA, are introduced. On this basis, the MATHLAB software is used to simulate the parameters that affect the performance of Turbo codes (that is, component codes, interleavers, deleters, iterations, decoding algorithms). The reasonable parameters to improve the performance of Turbo codes are given by analyzing the simulation results. Secondly, the Turbo code Interleaver is studied, and a new semi-regular Interleaver is proposed to overcome the shortcomings of the spiral Interleaver. The new interleaver uses random array to change the rank and column order of interleaved matrix from fixed to random. At the same time, the interleaved sequence is read out in positive spiral order, which makes the interleaved sequence have the characteristic of module 2, which increases the distance after interleaving of adjacent bits. The problem of uneven information bit protection caused by erasure matrix is solved. In order to make the semi-regular interleaver achieve the best working state, the column and column reset array is taken as the set to be solved, and the genetic algorithm is used to search for the best row and column reset array. Simulation results show that the performance of semi-regular Interleaver is better than that of packet Interleaver, Spiral Interleaver and Random Interleaver, which is close to S-type random Interleaver. After optimization, the performance of the semi-regular interleaver is kept in the optimal state, and the bit error rate (BER) gain under the same signal-to-noise ratio (SNR) is also obtained. Finally, in order to reduce the delay of iterative decoding and the average number of times of iterative decoding, the stopping criteria for iterative decoding of Turbo codes are studied, and two improved stopping criteria for iterative decoding are proposed, namely, the improved HDA criterion and the improved SDR criterion. The improved HDA criterion defines an unreliable metric based on external information. When the unreliable metric reaches the threshold, it is decided that the decoding of this frame cannot converge, and the decoding can be stopped in advance. The problem of failure of HDA criterion under low signal-to-noise ratio (SNR) is solved, and the improved SDR criterion uses the ratio of symbol difference number output of two successive iterations to replace the original decision function. The problem that the original criterion can not stop decoding when the frame is not convergent is solved. The simulation results show that the average number of iterations of the two improved criteria is significantly reduced than that of the original criterion without reducing the decoding performance.
【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】