基于LDPC码和喷泉码级联的研究

发布时间：2018-06-19 17:29

本文选题：喷泉码 + LT码　；参考：《杭州电子科技大学》2017年硕士论文

【摘要】：喷泉码作为一种较新颖的信道编码技术,凭借其无率和一致特性,以及无需反馈、编译码复杂度低、中断可续传的优点,近些年成为通信领域研究的热点之一。低密度奇偶校验(LDPC)码是一类基于稀疏校验矩阵结构的线性分组码,其检纠错能力强,性能可逼近Shannon限,译码复杂度较低,硬件实现简单,在第四代移动通信(4G)中有应用,且目前已被确定为第五代移动通信(5G)的长码编码方案。为了对抗信息传输过程中信道的干扰,确保通信的可靠性,尤其对于某些特殊环境,如深空通信、雷电等传输时延大、链路质量不佳的状况,本文提出一种喷泉码与LDPC码级联的编码方案,并探讨分析其性能与优势。本文首先介绍了数字喷泉码和LDPC码的理论知识,并对两类实用性喷泉码—LT码以及Raptor码展开了详细的研究,仿真分析了度分布对喷泉码的译码性能的影响以及几种常用的译码算法的性能及优缺点。此外对LDPC码的编译码算法的性能也进行了仿真比较。在基于LDPC码和喷泉码的研究的基础上,提出了两者的级联方案,为了综合评价该种级联方案,除AWGN信道外又建立了两种特殊的信道模型,一是在某段时间内发生了深衰落,二是在一段时间内随机出现多次短时信号中断,利用内码的检错删除机制将非删除信道模型等效为删除信道,使其适合喷泉码应用。对喷泉-LDPC级联码在不同的信道模型下进行了性能仿真,并尝试通过改变内码的码长、码率并加入交织来提高级联码的性能,此外还比较分析了喷泉-LDPC码级联方案和其它编码方案的表现。研究结果表明,Raptor码作为一种改进的LT码,在编译码复杂度和成功译码率方面均占据优势,作为级联码的外码更为合适;基于系统Raptor码和LDPC码的级联编码方案在AWGN信道以及构造的信道上均有不错的性能,相比单纯的LDPC码而言,编码增益有极大提高,其为信息的可靠传输提供了一种解决思路。最后阐述了Raptor-LDPC码在DSP上实现的具体过程,涉及到定点量化、矩阵存储、矩阵运算等,并探讨了不同量化方案和量化级数对Raptor-LDPC码误码性能的影响,并发现选择合适的量化方案可以将量化所造成的损失降到很低,减小对级联码性能的影响。通过理论研究和实际仿真可以发现,LDPC码和喷泉码级联不仅具有优异的码字性能,而且能保持较低的编译码复杂度,更为重要的是其抛弃了传统的自动反馈重传(ARQ)机制,可极大地缩短文件的传输延时,同时又确保成功译码率。它可用于深空通信、深海远程水声通信等环境,有效应对信道时变、传输时延长、误码率高及通信链路易中断的状况。
[Abstract]:Fountain code, as a novel channel coding technology, has become one of the hotspots in the field of communication in recent years because of its advantages of low coding and decoding complexity and low complexity of coding and decoding without feedback. Low density parity check (LDPC) codes are a class of linear block codes based on sparse parity check matrix structure. They have strong error-correcting ability, approximate Shannon limit, low decoding complexity, simple hardware implementation, and are used in the fourth generation mobile communication system (4G). And has been identified as the fifth generation mobile communication 5 G long code coding scheme. In order to counteract the interference of the channel in the process of information transmission and ensure the reliability of the communication, especially in some special environment, such as deep space communication, lightning and other transmission delay, the link quality is not good. This paper presents a concatenation scheme of fountain code and LDPC code, and discusses its performance and advantages. In this paper, the theoretical knowledge of digital fountain codes and LDPC codes is introduced, and two kinds of practical fountain codes-LT codes and Raptor codes are studied in detail. The effect of degree distribution on the decoding performance of fountain code is analyzed by simulation, and the performance, advantages and disadvantages of several common decoding algorithms are also analyzed. In addition, the performance of LDPC encoding and decoding algorithm is also simulated and compared. Based on the research of LDPC code and fountain code, the concatenation scheme of LDPC code and fountain code is proposed. In order to evaluate the concatenation scheme, two special channel models are established in addition to the AWGN channel. Secondly, several short time signal interruptions occur at random for a period of time. The non-deleted channel model is equivalent to the deleted channel by the error detection and deletion mechanism of the inner code, which makes it suitable for the application of fountain code. The performance of fountain LDPC concatenated code is simulated under different channel models, and the performance of concatenated code is improved by changing the code length, code rate and interleaving. In addition, the concatenation scheme of fountain-LDPC code and other coding schemes are compared and analyzed. The results show that Raptor code, as an improved LT code, is more suitable as a concatenated code in terms of complexity and rate of successful decoding. The concatenated coding scheme based on system Raptor code and LDPC code has good performance in AWGN channel and constructed channel. Compared with the simple LDPC code, the coding gain is greatly improved, which provides a solution for the reliable transmission of information. Finally, the realization process of Raptor-LDPC code on DSP is described, which involves fixed-point quantization, matrix storage, matrix operation, etc. The effects of different quantization schemes and quantization stages on the performance of Raptor-LDPC code are discussed. It is found that the loss caused by quantization can be reduced to a very low level and the effect on the performance of concatenated codes can be reduced by choosing the appropriate quantization scheme. Through theoretical research and practical simulation, it can be found that the concatenation of LDPC codes and fountain codes not only has excellent codeword performance, but also keeps low encoding and decoding complexity. More importantly, it abandons the traditional automatic feedback retransmission (ARQ) mechanism. It can greatly shorten the transmission delay of the file and ensure the successful decoding rate at the same time. It can be used in the environment of deep space communication, deep-sea long-distance underwater acoustic communication and so on. It can effectively deal with the condition that the channel is time-varying, the transmission time is prolonged, the bit error rate is high, and the communication link is easy to be interrupted.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN911.22

【参考文献】