Font Size: a A A

Application Research Of KDD On Analying Performance Data In Database Of The NMS

Posted on:2002-08-08Degree:DoctorType:Dissertation
Country:ChinaCandidate:C M LiFull Text:PDF
GTID:1118360032957072Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
The purpose of Network Management System(NMS) is to improve farthest network usability~ utilization rate, network performance,service quality and security so as to serve network's long-term plan. As network development, network management and maintenance is more important and more diffcult than before.By means of tradictional network management software,network manager can dynamically observe network device status and its location, however it is limited for network management. Tradictional network management software can't express devices's performance association,mutual affect of device on different location, and mutual affect of different protocol .These functional requirement have ont only realistic meaning for network management,but also are of importance for long-term network plan.On other hand, the NMS produces a large amount of data(including configuration data,performance data, fault data,alarm data) every day.These data are saved in database,we refer to it as the NMS database.There are many performance data of the network device in the NMS database. The performance data factually represent the network performance,and imply relationship between network devices. The knowledge discovery from the performance data have important significance for performance analysis and network maintenance.Hence, in order to research the network performance we must find new approach.KDD is the best method.The research in this paper focuses mainly on the following several aspects.oBased on Network Performance Data KDD Modeling in the NMS DatabaseKDD is the nontrivial process of identifying valid ,novel,potentially useful,and ultimately understandable knowledge from a large amount of data.This paper analyses the constitute of the NMS Database,and presents a method to construct transaction on the network management performance. Based on mib of snmp we clearly describe the attributes of network performance.Based on analysis of the traditional Al application on the NMS,and network-III-performance data in the NMS database,KDD model is given.oImplementation of DataWarehouse and OLAP based on Network Performance Data in the NMS DatabaseData Warehouse is data collection of subject -oriented, integrated, from different period.Subject. is the standard for classifying data.Every subject corresponds to one field to be analysed.For network performance data,we study how to efficiently organise data warehouse in this paper.OLAP make user obtain general , generalized knowledge by means of rolling up, drilling down,slicing,dicing,pivotting, etc.In this paper,the general knowledge categories can be obtained by use of OLAP and specific examples.This is important for network management performance data analysis.oConceptual Hierarchies Organization TechniqueConcepts in databases are organized into a partial order called conceptual hierarchies.Conceptual hierarchies play an important role in the knowledge discovery process because they specify backgroud or domain knowledge and may affect the discovery processing and the results.ln this paper,basic ideas about conceptual hierarchy and its manipulation are discussed.Based on the conceptual hierarchies in the NMS database,we present processing methodsoApplication Research of Multilevel Association Algorithm in the NMS DatabaseWe emphatically discuss performance data multilevel conceptual association problems on network management background, especially,the association between performance data and geographic location.After studying encoding multilevel association algorithms and background item constraint algorithms,an efficient mining multilevel association algorithm between performance data and geographic location is is efficient for mining useful association rules.
Keywords/Search Tags:Network Management, Network Performance, KDD, Datamining, Association Algorithm
PDF Full Text Request
Related items