基于数据挖掘的移动网络优化与运营技术研究

发布时间：2018-01-02 01:21

本文关键词：基于数据挖掘的移动网络优化与运营技术研究　出处：《北京邮电大学》2015年硕士论文　论文类型：学位论文

【摘要】：随着移动互联网业务的迅猛发展和智能终端的广泛渗透,移动通信网络的质量保障和业务运营方式受到了很大冲击。如何有效分析评估网络资源与业务需求的匹配程度,如何在资源受限的情况下动态调度资源以提升网络服务能力,如何深度挖掘用户行为模式以期提供更好的业务体验,这些问题都亟待解决。本文针对无线资源利用率分析与优化、客户业务偏好分析与推荐两个关键问题,深度挖掘网络运营过程中积累的历史数据,为网络侧无线资源配置优化、用户侧智能化业务推荐提供解决方案,从而提升网络的服务能力、服务质量和用户体验。论文研究内容分为两部分：1)评估网络无线资源利用率现状并提出一种资源动态调整框架和算法以提升资源利用效率。基于运营商的网管系统中提取小区资源配置数据和小区业务量数据,利用聚类、分类、相关性分析等数据挖掘技术,量化分析网络空闲资源量的时域和空域特征,证明网络无线资源与业务需求在时域和空域的不匹配特性。在此基础上提出了一种基于业务量预测的无线资源动态调整优化的框架,通过高斯加权K近邻算法预测业务量,通过优化算法动态调整网络资源配置,设计贪心算法求解优化问题,以提升资源利用效率并均衡网络负载。2)深度挖掘用户业务偏好,设计智能化业务推荐算法模型为用户推荐喜好的业务类型,为精细化营销奠定基础。首先通过深度包检测技术采集用户使用的数据业务,然后通过使用频次、使用流量等维度刻画用户业务偏好肖像,最后利用奇异值分解和最近邻算法设计混合的业务推荐算法,智能地为用户推荐喜好的业务类型。最后,对全文内容进行了归纳总结,分析了方案模型的待完善之处,并指出了后续的研究思路和方向。
[Abstract]:With the rapid development of mobile Internet penetration and intelligent terminal, to ensure the quality of the mobile communication network and business operations has received great impact. How to effectively analyze the matching degree of evaluation of cyber source with business needs, in the condition of limited resources dynamic scheduling resources to improve network service capacity, to the depth of user behavior pattern mining in order to provide better service experience, these problems should be solved.
According to the analysis and optimization of radio resource utilization, customer service preference analysis and recommendation of two key issues, the depth of mining historical data accumulated in the network operation, network side radio resource allocation optimization, user side intelligent business solutions, so as to enhance the service ability of the network, service quality and user experience. The content of this paper is divided into two parts: 1) wireless resource assessment network utilization situation and put forward a framework and algorithm of dynamic adjustment of resources to enhance the efficiency of resource utilization. The extraction of cell resource data and small business data management system based on the use of operators in clustering, classification, association analysis of data mining technology, quantization analysis of the temporal and spatial characteristics of the network free resources, network resources and business needs that wireless in matching the characteristics of time domain and space domain based on this. Presents a traffic prediction dynamic radio resource optimization framework based on the weighted K nearest neighbor algorithm to predict the Gauss business, through the optimization algorithm of dynamic adjustment of cyber source configuration, design a greedy optimization algorithm, to improve the efficiency of resource utilization and network load balancing.2) depth of mining user preference, intelligent design business model recommendation recommendation algorithm for user preferences for the type of business, to lay the foundation for the precise marketing. Firstly, through the deep packet inspection technology to collect user data, and then through the use of frequency, using the flow dimensions describe user preference image, finally using singular value decomposition and nearest neighbor algorithm to design hybrid recommendation algorithm. Intelligently recommend user preferences for business types.
Finally, the content of the full text is summarized, the improvement of the scheme model is analyzed, and the following research ideas and directions are pointed out.

【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN929.5;TP311.13

【共引文献】