【作者】 程；

【导师】 廖建新；

【作者基本信息】 北京邮电大学 ， 计算机科学与技术， 2008， 硕士

【摘要】 近年来,以彩铃业务为主要代表的电信增值业务的快速发展为运营商带来了丰厚的收益。然而固有的网络管理系统还停留在配置管理、性能管理、故障管理等孤立的网元层管理水平上,在提供以客户为中心的统一的业务运行质量管理上显得力不从心。本文首先从介绍智能网以及智能网网管系统的概念入手,分析了原有的网管系统存在的一些问题,提出了开发彩铃业务评价系统的必要性,然后通过彩铃评价模型和评价算法的介绍与设计,阐述了彩铃业务评价系统的基础,最后通过设计与实现该系统并检验评价结果证明了该系统的评价有效性。本文第一章分别介绍了固定智能网和移动智能网各自的特点,以及智能网网管系统的基本功能。然后重点分析了现有的网管系统在对彩铃业务等增值业务领域的管理上所存在的一些不足,进而提出了开发一套针对这些增值业务进行管理、分析、评价的系统的必要性。本文第二章和第三章分别介绍和设计了彩铃业务评价模型和评价算法。第二章根据电信管理论坛(TeleCommunication ManagementForum,TMF)提出的相关定义和服务质量(Quality of Service,QoS)白皮书标准,总结和归纳出了对于彩铃业务而言的一系列关键质量指标(Key Quality Indicator,KQI)和关键性能指标(Key PerformanceIndicator,KPI),并以此为基础构造出了彩铃业务评价系统的三层评价模型,然后结合东信北邮信息技术有限公司彩铃系统的特点,扩展成了五层评价模型。第三章重点介绍了层次分析法(AnalyticHierarchy Process,AHP),该方法已经被证明是解决同层次各要素之间相对重要性程度的有效方法。最后设计了KTK算法(KPI To KQI,KTK),该算法是满足客户需求的得分评价算法。从第四章开始,本文进入到系统的设计与实现及验证部分。首先对整个系统的架构做了概述,阐明了系统的逻辑结构和总体结构。然后将整个系统分为三个子系统来分别设计,其中KPI抽取子系统属于客户端(Client),部署在每台彩铃设备主机上,负责通过ODBC以及日志分析的方式抽取每台设备主机的KPI原始信息,而KPI元数据处理子系统和KQI计算与预警发送子系统同属一个进程,是服务器端(Server),部署在专用的彩铃业务评价系统主机上,负责进行日志信息的冲洗、格式转换、计算评价模型上每个节点的状态和得分以及预警条件的检查与发送。文中对上述三个子系统的概要设计、关键处理流程设计和关键类设计进行了详尽说明。最后通过一个实例说明了采用AHP算法来实现本系统与采用专家调查法和Delphi法相比,显著性更强,尤其在发生故障时,评价大幅度恶化,提醒维护人员及时处理。最后,本文在第六章重点分析了该系统目前还存在的不足以及值得改进的方向,为本系统的推广和下一个版本的设计提供了一些意见和建议。更多还原

【Abstract】 The rapid development of CRBT and other following value-added telecom services have brought operators huge benefits in recent years. But the original network management system (NMS) still focuses on network element layer management such as configuration management, performance management, alarm management and so on, which becomes insufficient in providing unified customer-oriented QoS management.By introducing the concepts of intelligent network (IN) and NMS for IN, the original NMS’s problems are exposed and the necessity of developing CRBT evaluating system is proposed in this thesis. Then the author designs CRBT’s evaluating model and evaluating algorithms, which become the basis of this system. Finally by implementing this system and verifing evaluating results, the system’s evaluating effectiveness is proved.In the first chapter of this article, we introduce the characteristics of fixed intelligent network and mobile intelligent network as well as the functions of intelligent network managenment system (INMS). Then disadvantages of current INMS’s management on value-added services such as CRBT are analysed carefully, which makes it quite necessary to develop a system to manage, analyse and evaluate value-added services.The evaluating model and evaluating algorithms of CRBT are introduced and designed respectively in chapter 2 and 3 of this thesis. At first, a series of KQIs (Key Quality Indicator) and KPIs (Key Performance Indicator) for CRBT services are concluded based on TMF’s (Telecommunications Management Forum) related definitions and white paper of QoS (Quality of Service). Then the second chapter brings a 3-layer evaluating model for CRBT, which is constructed based on those indicators and then expands into 5-layer model according to the features of EBUPT’s (Eastcom-bupt technology co., ltd) CRBT system. At last we describe AHP algorithm, which is proved to be an effective algorithm to determine all factors’ relative degree of importantance in a same layer and carefully design KTK algorithm, which is a scoring algorithm that can satisfy custermor’s requirements in chapter 3.The system’s design, implementation and results verification start from the fourth chapter. First of all, we summrise the entire system’s architecture and describe the system’s logical and overall structure. According to this article, the whole system consists of 3 subsystems: KPI extracting subsystem, KQI metadata handling subsystem and KQI calculating and alarm handling subsystem. The first subsystem, whose duty is to extract every host’s original KPI information through ODBC as well as log analysis, belongs to Client and deploies on every host that run CRBT services. The other two are in the same process and belong to Server which deploies on specific host that run CRBT evaluating system. Their duty is to process log format conversion, to calculate every node’s score and status in the evaluating model as well as inspecting and sending early warning. Then the above three subsystems’ overview design, key process and key class design are described in detail in chapter 4. Finally, the fifth chapter lists an example to demonstrate that, compared to Experts algorithm and Delphi algorithm, AHP algorithm has more significant results. Especially when some fault happens, the system’s score will be reduced rapidly to remind maintenance staff to take measures in time.However, the system still has some disadvantages. And in the last chapter, we analyse those disadvantages and provide some directions worthy of improving as well as some useful recommendations and suggestions for the system’s promotion and the design of next version.更多还原