Research On Application Of Data Acquisition And Data Mining In Mobile Communication Network

With the development of communication technologies and the increase of mobile terminals,the mobile communication network has accumulated amounts of data.Meanwhile,the complexity and mobility of the mobile communication network have also continuously increased,which causes the problem that network quality did not match the network scale,and the network cannot achieve optimal operation.In this thesis the optimized troubleshooting of VoLTE voice quality factors in mobile communication networks is taken as an example,and data collection and data mining technologies are used to analyze and research the massive data in mobile networks.The research is mainly in the following three aspects:(1)The data analysis process for mobile communication networks is established based on relevant theories of data acquisition and mining technology applied to mobile communication networks.(2)The reason of VoLTE voice quality problem is analyzed based on clustering algorithm.MOS values are used to characterize voice quality.Then,pre-processing and statistical analysis of collected VoLTE voice quality data are conducted to find the direct causes(voice coding rate,packet loss,delay,jitter)of MOS values and 12 underlying causes which lead to the direct causes(alarm,interference,capacity,coverage,poor quality,RRC reconstruction,uplink power limited,frequent handover,eSRVCC handover,parameter setting,transmission,terminal and core network).The influencing factors of the MOS values are clustered and analyzed by using hierarchical clustering algorithm to get the correspondence between the direct cause and the underlying cause.(3)Troubleshooting and optimization methods for underlying causes of VoLTE voice quality issues are proposed.Four underlying reasoning finding decision trees are built by using ID3 decision tree mining,which aimed at the four direct causes of the VoLTE voice quality problem.Then four underlying reason optimization order decision trees are established by using the CART decision tree mining.After that,the optimal investigation order and optimization sequence of the underlying causes of VoLTE voice quality problems can be determined.An underlying causes troubleshooting method of the VoLTE voice quality problem is proposed in this thesis.Hierarchical clustering algorithm is used to get correspondence between the direct cause and the underlying cause.ID3 decision tree and CART decision tree are used to identify and optimize the underlying causes of VoLTE voice quality and make optimal decisions based on the actual situation.VoLTE voice quality problems could be solve quickly and efficiently by using this troubleshooting method.