高速turbo译码器迭代终止策略研究

发布时间：2018-03-14 17:00

本文选题：长期演进系统　切入点：turbo码　出处：《华中科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：turbo码是一种性能优越的信道编码技术，3GPP长期演进LTE系统和LTE-Advanced系统中的信道编码均采用了turbo码。LTE-A系统标准中要求下行峰值速率1Gbit/s，上行峰值速率500Mbit/s。为符合LTE-A系统1Gbit/s的高传输速率的要求，需要对传统的turbo编译码器进行改进，设计和验证符合未来无线通信系统中高传输速率要求的新型译码算法架构。迭代终止策略是提高译码速度的关键技术之一，能够在误码性能损失较小的情况下大幅减少turbo译码过程中平均迭代次数，提高迭代译码速度。本文主要针对turbo码译码迭代停止准则进行研究和改进，对常见的迭代停止准则在高速传输应用场景下进行仿真分析，筛选出迭代速度最高的S3准则，，并针对S3准则提出了一种新的改进算法，利用软信息参量的统计值作为度量来判断是否应停止迭代过程。本文还新提出了另一种基于余弦相似度的算法，采用两个SISO译码器输出的外信息序列的余弦相似度值作为度量，来判断迭代过程是否需要终止。仿真结果表明，与已有迭代停止算法相比，两种新算法均大幅降低了平均迭代次数，提高了turbo码译码器的译码速度；而在误码性能方面只有很小的损失或几乎没有损失。S3改进算法在高比特信噪比环境下对译码速度的提高十分明显；而余弦相似度算法在低比特信噪比环境下节省迭代次数效果显著。本文提出的两种新算法均可根据实际应用环境用于设计高速turbo码译码器，以满足对于turbo码高速译码的需求。
[Abstract]:Turbo code is a channel coding technique with excellent performance. Both 3GPP long-evolving LTE system and LTE-Advanced system adopt turbo code. LTE-A system standard requires downlink peak rate of 1Gbit / s and uplink peak rate of 500Mbit / s. High transmission rate requirements, It is necessary to improve the traditional turbo codec and design and verify a new decoding algorithm architecture that meets the requirements of high transmission rate in future wireless communication systems. Iterative termination strategy is one of the key techniques to improve the decoding speed. It can greatly reduce the average number of iterations in the turbo decoding process when the performance loss is small. In this paper, the iterative stopping criteria of turbo codes are studied and improved. The common iterative stopping criteria are simulated and analyzed in high-speed transmission scenarios, and the S3 criterion, which has the highest iterative speed, is selected. A new improved algorithm is proposed for S3 criterion, which uses the statistical value of soft information parameters as a measure to judge whether the iterative process should be stopped. Another algorithm based on cosine similarity is also proposed in this paper. The cosine similarity values of the external information sequences outputted by two SISO decoders are used as measurements to determine whether the iterative process needs to be terminated. The simulation results show that the proposed method is more efficient than the existing iterative stopping algorithms. The two new algorithms greatly reduce the average number of iterations and improve the decoding speed of the turbo decoder. However, in the aspect of error performance, the improved S3 algorithm has little or no loss, which can improve the decoding speed obviously in the environment of high bit signal-to-noise ratio (SNR). CoSine similarity algorithm can save iteration times in low bit signal-to-noise ratio (SNR) environment. The two new algorithms proposed in this paper can be used to design high speed turbo decoder according to the actual application environment to meet the needs of high speed decoding of turbo codes.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.22

【相似文献】