面向5G的稀疏码分多址技术研究

发布时间：2018-06-28 19:57

本文选题：5G + 多用户检测　；参考：《电子科技大学》2017年硕士论文

【摘要】：稀疏码分多址(Sparse Code Multiple Access,SCMA)技术是一种面向第五代移动通信(Fifth Generation,5G)系统的新型非正交多址技术。相对于传统的正交多址技术,其拥有更高的频谱利用率和更大的网络承载能力,因而成为未来5G空口技术的研究热点。与低密度信号(Low Density Signature,LDS)不同,在发送端,SCMA把星座点映射和扩频相结合,直接将二进制比特数据映射成多维复码字。多维复星座图带来的成型增益使SCMA性能相比于LDS有很大的提升。在接收端,由于码字稀疏性,SCMA接收机可以利用消息传递算法(Message Passing Algorithm,MPA)进行多用户检测(Multi-user Detection,MUD),即使在系统严重过载的情况下也能获得很好的性能。MPA算法是一种基于因子图求边缘概率分布的迭代算法,该算法中外信息在变量节点(Variable Node,VN)和函数节点(Function Node,FN)之间不断的传递,最后获得一个稳定的概率分布作为判决量,最终最优的判决量对应的码字即为判决输出结果。相对于联合最优最大后验概率(Maximum-a-posteriori,MAP),MPA是一种次优的方法,但是MPA利用了码本的稀疏性,极大地降低了多用户检测的复杂度。虽然MPA算法的复杂度相对于最优的MAP有所降低,但是在系统严重过载或码本尺寸过大的情况下,硬件实现仍然很困难。目前降低MPA复杂度的方法有很多,如加权消息传递算法(Weighted Message Passing Algorithm,WMPA),串行消息传递算法(Shuffled Message Passing Algorithm,SMPA),低映射点的码本设计等,这些方法归结起来可以分为两类:一类是改进算法本身,一类是优化码本。本文主要通过改进MPA算法来降低复杂度,针对上行和下行链路分别提出了改进方案。本文的主要工作如下:(1)在上行SCMA系统中,提出了一种基于门限的低复杂度MPA算法。其基本思路是:在每次迭代完成后,利用门限来选择置信度高的码字进行译码。由于码本具有串行干扰消除(Success Interference Cancellation,SIC)特性,相互叠加的用户中能量大的用户会很容易被检测出来。与原始MPA相比,在信噪比较高的情况下,所提算法中绝大多数用户不需要迭代到最大迭代步数就被译码,整个译码过程中用户的平均迭代步数减少,系统复杂度降低。同时,在门限设置合适的情况下,可以获得与原始MPA相当的系统性能。(2)在SMPA基础上,提出了一种基于压缩码本的改进SMPA算法。该算法的主要思想是:利用SMPA的快速收敛特性,及时对发送概率极低的码字进行舍弃,从而缩小后续迭代过程中单个消息更新所需的搜索范围,进而达到进一步降低SMPA复杂度的目的。(3)针对下行SCMA系统,提出了一种改进的MPA算法。该算法从单个用户角度考虑,只关注与该用户相关频点在内的部分频点,在因子图中消息仅沿着少量的边进行传递。相对于原始的多用户联合检测MPA算法而言,每次迭代的复杂度减少,同时,相对于不迭代的MPA所提算法的性能有所提升,因为在迭代后码字概率分布优于初始化的概率分布。
[Abstract]:Sparse Code Multiple Access (SCMA) technology is a new non orthogonal multiple access technology for the fifth generation Fifth Generation (5G) system. Compared with the traditional orthogonal multiple access technology, it has higher spectrum utilization and larger network carrying capacity, thus becoming the research heat of the future 5G empty technology. Different from the low density signal (Low Density Signature, LDS), at the sending end, SCMA combines the constellation point mapping and the spread spectrum to map the binary bit data directly into a multidimensional complex code. The multidimensional complex constellation can improve the SCMA performance greatly compared to the LDS. At the receiver, the SCMA receiver can be a SCMA receiver due to the codeword sparsity. Message Passing Algorithm (MPA) is used to carry out multiuser detection (Multi-user Detection, MUD), even if the system is seriously overloaded, the.MPA algorithm is an iterative algorithm based on the factor graph to calculate the edge probability distribution. The information in this algorithm is in the variable node (Variable Node, VN). The continuous transmission between the function node (Function Node, FN), and finally a stable probability distribution as the judgment, the final optimal decision word is the result of the decision output. Relative to the joint optimal maximum posterior probability (Maximum-a-posteriori, MAP), MPA is a suboptimal method, but MPA uses the thinning of the codebook. The complexity of multiuser detection greatly reduces the complexity of multiuser detection. Although the complexity of the MPA algorithm is lower than the optimal MAP, the hardware implementation is still difficult when the system is seriously overloaded or the size of the codebook is too large. At present, there are many ways to reduce the MPA complexity, such as the weighted message transfer algorithm (Weighted Message Passing). Algorithm, WMPA), the serial message transfer algorithm (Shuffled Message Passing Algorithm, SMPA), the codebook design of low mapping points, and so on. These methods can be divided into two categories: one is to improve the algorithm itself and the other is the optimization of the codebook. This paper is mainly to reduce the complexity by improving the MPA algorithm, and propose to the uplink and downlink respectively. The main work of this paper is as follows: (1) a low complexity MPA algorithm based on threshold is proposed in the uplink SCMA system. The basic idea is that after each iteration is completed, the threshold is used to select a codeword with high confidence level. Because the codebook has a series line interference cancellation (Success Interference Cancellation, SIC) Compared with the original MPA, the majority of the users in the proposed algorithm do not need to be iterated to the maximum number of iterations, and the average iteration number of the user is reduced and the system complexity is reduced in the whole decoding process. Under the appropriate circumstances, the performance of the system equivalent to the original MPA can be obtained. (2) an improved SMPA algorithm based on the compression codebook is proposed on the basis of SMPA. The main idea of this algorithm is to use the fast convergence property of SMPA, to abandon the very low transmission probability in time, and to reduce the single message in the subsequent iteration process. Update the required search range to further reduce the SMPA complexity. (3) an improved MPA algorithm is proposed for downlink SCMA systems. This algorithm is considered from a single user angle, focusing only on the frequency points associated with the user, and the message is transmitted only along a small edge in the factor graph. As far as the initial multiuser joint detection MPA algorithm is concerned, the complexity of each iteration is reduced, while the performance of the algorithm compared with the non iterative MPA is improved because the probability distribution of the codeword probability is better than that of the initialized probability distribution after the iteration.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN929.5

【参考文献】