【作者】 张帅；

【导师】 宋荣方；

【作者基本信息】 南京邮电大学 ， 电子与通信工程（专业学位）， 2022， 硕士

【摘要】 无线通信技术的发展,给人类社会生活带来了极大的便利,每个人的社会活动都随时进行着信息传递,个人信息传输的效率和质量应是一个需要重点研究的问题。尤其在5G,6G等高频通信技术下,会有更多的终端接入互联网,需要传输的数据将是以前不可想象的。因此,对于无线通信系统信息传递的全过程,各个阶段的研究必须进行客观深入的开展,才能为人类生活提供更好的保障。在非正交多址(Non-Orthogonal Multiple Access,NOMA)系统中,要对用户的数据进行精准的传输,那么首先要对用户进行区分,也就是用户聚类,以便基站能够准确识别每个用户。在接收端用户接收到信息时,能够准确的进行解调,获得自己想要的信息,这是第一步。在对用户进行分簇之后,要考虑的问题就是对不同的用户根据通信规则进行不同的功率分配,这也是对用户信息进行解调的关键依据。因此,对用户的分簇与功率分配是研究的关键问题。本文也将对这两个问题进行研究。论文的主要研究内容如下:首先,对数据聚类算法进行研究,提出了一种基于无监督学习的用户分簇算法,针对通信系统模型,建立系统容量最大化的优化模型,利用K-Means聚类算法以及改进算法的特点,提出一种新的用户分簇算法,针对用户特征,对用户进行合理分类。在多用户的环境下,通过仿真分析,发现该算法比以往采用的穷举搜索算法的复杂度明显降低,收敛速度也较之前快,系统容量得到提升。其次,针对用户功率分配问题,得益于进化算法的迭代,更新求优的特点,本文比较了遗传算法和粒子群优化算法,发现粒子群优化算法效果更好。于是本文提出利用粒子群优化算法对用户功率进行分配。经过仿真分析发现,本文采用分配算法与现有的算法相比,该算法达到最优值的速度较快,能有效提高系统容量,稳定性也更高。更多还原

【Abstract】 The development of wireless communication technology has brought great convenience to human social life,and everyone’s social activities carry out information transmission at any time,and the efficiency and quality of personal information transmission should be a problem that needs to be studied in focus.Especially in 5G,6G and other high frequency communication technologies,there will be more terminals to access the Internet,and the data to be transmitted will be unimaginable before.Therefore,for the whole process of information transmission of wireless communication system,the research of each stage must be carried out objectively and deeply in order to provide a better guarantee for human life.In Non-Orthogonal Multiple Access(NOMA)system,to transmit the user’s data accurately,then the user must first be distinguished,that is,the user clustering,so that the base station can accurately identify each user.At the receiving end when users receive the information,they can accurately demodulate and get the information they want,which is the first step.After the user clustering,the issue to be considered is the different power allocation to different users according to the communication rules,which is also the key basis for the demodulation of user information.Therefore,the clustering and power allocation of users are the key issues to be studied.In this paper,these two issues will also be studied.The main research of the paper is as follows.First,the data clustering algorithm is studied,and a user clustering algorithm based on unsupervised learning is proposed.An optimization model for maximizing system capacity is established for the communication system model,and a new user clustering algorithm is proposed using the K-Means clustering algorithm and the features of the improved algorithm to reasonably classify users with respect to their characteristics.In a multi-user environment,through simulation analysis,it is found that the algorithm has significantly lower complexity and faster convergence than the previously used exhaustive search algorithm,and the system capacity is improved.Second,for the user power allocation problem,thanks to the iteration of evolutionary algorithms,the characteristics of updating for optimization,this paper compares the genetic algorithm and the particle swarm optimization algorithm,and finds that the particle swarm optimization algorithm works better.So this paper proposes to use the particle swarm optimization algorithm to allocate the user power.After simulation analysis,it is found that the allocation algorithm used in this paper is faster in reaching the optimal value,can effectively improve the system capacity and has higher stability compared with the existing algorithms.更多还原