高维度SCMA码本设计与低复杂度译码算法研究

发布时间：2018-06-17 18:18

本文选题：SCMA + 码本设计　；参考：《西南交通大学》2017年硕士论文

【摘要】：稀疏码分多址(SCMA)技术是新一代5G通信技术的候选多址技术方案之一,该技术下,各个用户的信号在资源上进行非正交叠加,从而复用层数大于扩频因子,在资源数相同的情况下,系统能接入更多的用户,使资源复用率大大提高。在LDS MA(Low Density Signature Multiple Access)的基础上,SCMA将调制功能模块和扩频功能模块组合,将比特流直接映射成多维码字。与LDS中固定使用QAM符号的策略相比,SCMA系统通过采用更加复杂精细的星座图,来获得更大的编码与赋形增益,从而使得其系统性能优于LDS。因此SCMA系统中的码本设计是该研究方向的核心问题之一。研究表明,目前对SCMA码本的研究多限于较低维度码本,而对高维度的SCMA码本的性能和检测复杂度的讨论较少。而事实上SCMA系统正是为接入海量用户提供便利,因此研究高维度的SCMA码本并分析其性能显得很有意义。本论文研究了性能更优的基于"Latin" rectangular的SCMA新型码本构造方法,构造了高维度下的用户码本,并在高斯信道和瑞利衰落信道中对系统性能进行了研究。高维度的码本与优化的MPA检测算法进行结合,在提高系统鲁棒性的同时保证了较低的译码复杂度,因此拥有更优异的系统综合性能表现。在SCMA系统中,每个用户都有一个特定的码本,由码本和映射矩阵编码而成的码字具有稀疏性,在解码端采用具有良好性能的消息传递算法进行检测。传统的ML算法需要进行穷举搜索,如此导致运算复杂度很高,不利于实践中推广使用。因此SCMA系统借鉴了在LDPC码中大量使用的MPA解码算法,大大降低了多用户系统的解码复杂度。但是即便如此,在迭代次数过多,用户数量大增,以及追求更大的系统分集增益的这些场景下,MPA算法复杂度也将急剧增加。因此本论文深入研究了解码端的MPA算法。针对现有MPA算法复杂度较高的不足,对MPA算法进行了优化,研究了低译码复杂度的算法,并兼顾讨论了算法对系统误码性能的影响。
[Abstract]:The sparse code division multiple access (SCMA) technology is one of the candidate multiple access schemes for the new generation 5G communication technology. Under this technology, the signals of each user are superposed on the resources, so that the number of multiplexed layers is greater than the spread spectrum factor. In the case of the same number of resources, the system can access more users, so that the reuse rate of resources is greatly improved. On the basis of low density signature multiple access, SCMA combines modulation function module with spread spectrum function module, and directly maps bit stream to multidimensional codeword. Compared with the strategy of using QAM symbols fixed in LDS, SCMA systems can obtain greater coding and shape gain by using more complex and fine constellation diagrams, thus making the system performance better than LDSs. Therefore, the codebook design in SCMA system is one of the core problems in the research direction. The research shows that most of the researches on SCMA codebook are limited to the lower dimension codebook, but the performance and detection complexity of the high dimensional SCMA codebook are less discussed. In fact, the SCMA system is convenient to access a large number of users, so it is meaningful to study the high-dimensional SCMA codebook and analyze its performance. In this paper, a novel codebook construction method based on "Latin" rectangular is studied, and the system performance in Gao Si channel and Rayleigh fading channel is studied in the Gao Si channel and Rayleigh fading channel. The combination of high-dimensional codebook and optimized MPA detection algorithm can improve the robustness of the system and ensure lower decoding complexity, so it has better performance. In the SCMA system, each user has a specific codebook. The codewords encoded by the codebook and the mapping matrix are sparse. At the decoding end, a message passing algorithm with good performance is used to detect the codewords. The traditional ML algorithm needs exhaustive search, which leads to high computational complexity, which is not conducive to popularization in practice. Therefore the SCMA system draws lessons from the MPA decoding algorithm which is widely used in LDPC codes and greatly reduces the decoding complexity of multi-user systems. But even if the number of iterations is too many, the number of users will increase greatly, and the complexity of the MPA algorithm will increase sharply in these scenarios where the system diversity gain is larger. Therefore, this paper deeply studies the MPA algorithm on the decoding side. In view of the high complexity of the existing MPA algorithm, the MPA algorithm is optimized, the algorithm with low decoding complexity is studied, and the influence of the algorithm on the system error performance is discussed.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN929.5

【参考文献】