| The extensive use of smart devices and rapid development of mobile data service bring serious challenge to efficient operation of mobile network. How to accurately access the network resources demand in view of the trend of cell functional specialization? How to accurately determine whether current resource allocation scheme can meet the demand of business development? How to evaluate the effectiveness of a certain demand planning? These are some of the most urgent issues to tackle.This thesis, starting from the consideration of cell clustering and subscriber behavior analysis, aims at providing technical basis and guidance for accurate network optimization with utilization of service characteristic mining.First, Appropriate feature vector is constructed in the purpose of cell representation with the utilization of both geographical attributes and service attributes collected from network infrastructure using DPI. Time complexity, operability and practicality are jointly considered in the selection of clustering method. With the establishment of clustering label, service traits and coverage traits of each cluster are analyzed which is expected to provide valuable guidance to network optimization. In order to enhance the applicability of our work, detailed characteristic label is defined with dual explanation. Possible approach to apply clustering result in the deployment of network resource is observed with its advantage analyzed. Moreover, we put forward a method to accurately measure the user perceptive rate with the interference of subscriber behavior and confirm the effectiveness of this method in present network. Then detailed service preference analysis, service traffic analysis and service arrival distribution analysis is conducted as a foundation of mobile network precise management.Finally, a conclusion is presented to outline the imperfection of current scheme and indicate vision and direction for future work. |
PDF Full Text Request