蜂窝网络下基于上下文信息的无线资源分配策略研究

发布时间：2018-05-15 08:07

本文选题：蜂窝网络 + 密集小站　；参考：《中国科学技术大学》2017年硕士论文

【摘要】：随着智能终端的迅速普及和移动互联网的高速发展,用户的数据业务需求和通信场景日益多样化,移动数据呈爆发式增长。为应对海量数据的高速增长以及实现对新场景的支持,第五代移动通信系统(Fifth-Generation Mobile Communication System,5G)提出系统容量实现1000倍增长的目标,大规模天线、超密集组网以及终端直通(Device-to-Device,D2D)等技术均被视为5G通信系统的关键技术。另一方面,数据量的激增虽然增加了处理开销,但也带来了潜在价值。通过对特定场景上下文信息中隐含信息的挖掘,对用户行为进行分析以对其未来请求做出预测,并将其用于5G新场景下的无线资源分配,以进一步提升系统容量具有重要的研究意义。在蜂窝网络中,根据5G通信系统的技术场景,可以分为连续广域覆盖、热点高容量、低时延高可靠和低功耗大连接等几类。由于不同场景下的业务需求以及对上下文信息的获取、利用不尽相同,进而对蜂窝网络中的无线资源分配造成不同的影响。因此,本文将聚焦密集小站网络和蜂窝D2D网络,研究这两个具体场景下,基于上下文信息的无线资源分配策略。针对密集小站网络场景,由于该场景下小站回程链路容量受限,通常考虑在小站上部署缓存以避免流性文件重复传输。因为小站缓存的加入,用户的传输速率除与物理信道条件有关,还与其请求文件在缓存中的命中率及回程链路容量有关。基于此,本文根据用户历史请求记录等上下文信息,利用基于用户的Top N协作滤波推荐系统初始化用户缓存,进而提出一种在满足用户间吞吐量比例公平性约束下,最大化系统吞吐量的无线资源分配策略。由于形成的优化问题涉及用户归属和缓存决策两组变量且非凸,难以直接求解。因此本文提出依次迭代更新两组变量的用户归属更新算法和缓存决策更新算法,然后再进行满足用户间吞吐量比例约束的无线资源分配。仿真结果表明所提算法能带来缓存命中率的提升,相较两种对比算法在系统吞吐量上有明显增益。针对蜂窝D2D网络场景,由于频谱资源稀缺,因而该场景下D2D对需要与直接与基站相连的蜂窝用户共享频谱资源,这会在二者间产生干扰导致蜂窝用户服务质量下降。而在现实生活中用户通常倾向于与自己亲密的人共享资源。基于此,本文利用社交网络中的基本信息和交互信息等上下文信息,定义并计算用户间的亲密度,然后联合考虑物理信道条件和用户间亲密度,提出一种在满足蜂窝用户和D2D对服务质量要求前提下的资源分配策略以最大化系统吞吐量。由于该优化问题为混合整数非线性规划问题,难以直接求解。因此本文通过将其分解为接入控制、功率分配以及蜂窝用户与D2D对匹配三个子问题,并依次求解子问题的方式来求解原问题。仿真结果表明所提算法在系统吞吐量上明显优于四种对比算法,能够更好的适用于频谱资源稀缺的蜂窝D2D网络场景。
[Abstract]:With the rapid popularization of intelligent terminals and the rapid development of mobile Internet, the user's data business requirements and communication scenarios are increasingly diversified, and mobile data are growing rapidly. In response to the rapid growth of mass data and the support for the implementation of the new scene, the fifth generation mobile communication system (Fifth-Generation Mobile Communication System, 5G) the goal of 1000 times growth of system capacity is proposed. Large scale antenna, ultra dense network and Device-to-Device (D2D) are considered as the key technology of 5G communication system. On the other hand, the increase of data amount, although increasing processing overhead, also brings potential value. The mining of hidden information, the analysis of user behavior to predict the future request, and use it to allocate the wireless resources in the new 5G scene, to further improve the capacity of the system has important research significance. In the cellular network, according to the technical scene of the 5G communication system, it can be divided into continuous wide coverage, hot and high capacity, low level. The time delay high reliability and low power consumption large connection and so on. Due to the different scene business needs and the access to the context information, the use of the wireless resource allocation in the cellular network is different. Therefore, this paper will focus on the dense small station network and the bee nest D2D network, based on the two specific scenarios, based on this A wireless resource allocation strategy for context information. For dense small station network scenarios, because the backlink capacity of a small station is limited in this scenario, it is usually considered to deploy caching on a small station to avoid the repeated transmission of the stream file. The hit rate in the cache is related to the return link capacity. Based on this, the user's Top N collaborative filtering recommendation system is used to initialize the user's cache according to the user's historical request record and so on, and then a wireless resource partition that maximizes the system throughput is proposed to satisfy the fairness of the inter user throughput ratio. As a result of the optimization problem involved in the user ownership and the caching decision, two groups of variables are not convex, and it is difficult to solve the problem directly. Therefore, this paper proposes a subscriber ascription updating algorithm and a cache decision updating algorithm which updates the two groups of variables in turn, and then carries out the wireless resource allocation that satisfies the ratio constraint of the throughput between users. It shows that the proposed algorithm can bring the increase of cache hit rate, compared with the two contrast algorithms, there is a significant gain in the system throughput. For the cellular D2D network scene, because of the scarcity of spectrum resources, the D2D needs to share the spectrum resources with the cellular users directly connected with the base station, which will cause interference among the two and cause the cellular use. In real life, users tend to share resources with their intimate people. Based on this, this paper uses the basic information and interactive information in the social network to define and calculate the intimacy between users, and then jointly consider the physical and channel conditions and the intimacy of users, and propose a kind of information. The resource allocation strategy of the foot cellular users and the D2D is to maximize the system throughput. Because the optimization problem is a mixed integer nonlinear programming problem, it is difficult to solve the problem directly. Therefore, this paper decomposes it into access control, power allocation, and cellular users and D2D pairs to match three sub problems, and solve the problem in turn. The simulation results show that the proposed algorithm is better than the four contrast algorithms in the system throughput, and can be better suitable for the cellular D2D network scene with the scarcity of spectrum resources.

【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN929.5

【相似文献】