【作者】 梁雨霏；

【导师】 李书芳；

【作者基本信息】 北京邮电大学 ， 通信与信息系统， 2013， 硕士

【摘要】 网络优化是一项贯穿网络运营始终的工作。通过采集OMC统计数据、DT/CQT数据、用户投诉数据和告警信息,并对这些数据加以分析,可以发现网络故障并加以排除,从而实现保证网络性能、提高用户满意度这一目标。然而,随着移动通信事业的快速发展,人们对于业务种类及质量的要求越来越高。与此同时,各地区不断增多的用户群体数也加重了网络的负荷。在这样的背景下,传统的网络维护与优化工作遭遇了新的挑战。为了解决上述问题,我们期望在传统的网络优化工作中引入先进技术,从而更加可靠且高效地完成网优工作。随着信息化社会的发展,各行各业都存储了海量数据。数据挖掘技术可以充分利用这些数据并从中寻找到有用信息。它具有如关联分析、预测、分类和聚类等多种功能。我们可以利用这些功能处理海量数据,找出数据间所包含的潜在信息。论文首先详细介绍了传统网络优化的流程和工作中涉及的统计指标和无线参数,并说明如何利用这些数据来分析网络当前性能、定位网络问题。而后说明传统网络优化的缺点和不足,并对数据挖掘技术应用于网络优化工作进行了可行性分析。笔者利用数据挖掘技术中的回归分析功能来预测话务量的发展趋势。文中说明了话务预测的具体流程,并介绍了真实数据的处理方法,对现有数据进行建模,最终通过模型预测短期内的话务变化。接下来,笔者讨论了如何发现某基站下性能较差的小区。文中首先说明传统方式的滞后性,并提出用数据挖掘技术中的聚类功能对小区性能进行划分,找出其中质量较差的小区。该方法高效、及时,具有较强的推广价值。而后,笔者研究如何利用数据挖掘中的分类的功能分析网络故障原因。文中结合真实案例进行说明,对处理结果进行了详尽的分析。论文最后分析了网络优化工作中可应用该技术的切入点,对数据挖掘技术在网络优化工作中的应用进行了展望。更多还原

【Abstract】 Network optimization is an important category of task which impacts every aspects of network operation. By collecting and analyzing OMC statistic data, DT/CQT data as well as user complaints data and alarm information, network malfunctions can be identified and cleared in time such that network performance can be guaranteed and user’s satisfaction can be improved.However, with the rapid development of the mobile communication industry, user’s expectation for business types and service quality become increasingly higher. At the same time, the expansion of user population in different areas can also overburden the network. Under such background, the traditional network maintenance and optimization encountered new challenges. In order to address these problems, we expect to introduce new advanced technologies to traditional network optimization so that we can tackle the task more reliably and efficiently.As a leading-edge technology, data mining originates from the development of data warehouse, which makes full use of the massive data and extracts more useful information in these data. This technology is associated with multiple functions such as correlation analysis, forecasting, classification and clustering. By using these functions, we are able to discover the hiddern knowledge behind the huge amounts of data.This dissertation briefly introduces the process of traditional network optimization and analyzes the statistic data and wireless parameter deeply. How to use the data to analyze performance of the network and pin point problems in the network will be explained in detail. This dissertation also discusses the weakness and shortage in traditional network optimization, and conducts a feasibility analysis.The author uses the technology of regression analysis in data mining to predict traffic trends. This dissertation describes the flow of forecasting traffic and elaborates the processing for the existed data, building a modelbased on these data, which is used to predict the short-term traffic. This paper puts forward a new way to find inferior cell in base stations, which is more efficient and timely, and will be compared with the traditional one. At last, the writer analyzes the root-causes of network problems with data mining technology. Associated with a practical case, this paper gives a detailed analysis.At last, the author presents some discussionon in which fields data mining can be used along with prospect of network optimization with data mining in the future.更多还原