Design And Implementation Of Performance Analysis And Early Warning System In Mobile Core Network

In recent years, with the development of mobile communication network, mobile network structure is becoming more and more complex. Users ask for higher quality of network. Thus the mobile network maintenance is becoming more and more important. The traditional reactive fault handling has been unable to meet the requirements of operators for the network quality. How to control the real-time network running status, how to give the early warning for hidden network performance trouble, and how to move forward the user complaints are significant issues to be solved for operators at present.Aiming at the MSC Pool mobile network structure of the branch company of China Mobiles in a certain city, this paper designs a set of the performance analysis and early warning system in mobile core network. The system is used to monitor changes of performance in mobile core network, give early warnings for abnormal performance index, so that network maintenance workers can analyze and process as soon as possible, reducing the network failures. The performance analysis and early warning system in mobile core network is mainly composed of data acquisition module, KPI calculation module, early warning and prediction module, mail module and data presentation module. Data acquisition module downloads data files regularly from mobile core network equipment MSC Server through FTP, and save them to local directory. KPI calculation module reads the files which are downloaded completely, calculates KPI according to the KPI formula configuration, and saves the KPI value to database. KPI formulas can be configured freely by the user. Early warning and prediction module is used to compare real-time KPI value to warning threshold. Early warning is gotten when real-time KPI value is out of the threshold. Early warnings are aggregated to main warning, as prediction, which are used to reflect the overall network condition. This system uses two levels of early warning, the high-level early warning and the low level, which is based on static threshold and dynamic threshold as the warning threshold. Static threshold is configured by the user, and dynamic threshold uses time series decomposition and K-medoids to calculate. Mail module is used to send network condition to the appropriate staff by mail, so that they can analyze and process in time. Data presentation module is for human-computer interaction. It is mainly for KPI query, early warning and main warning query, KPI configuration and system configuration. The system development platform is Visual Studio 2008, and the development language is VC++. The database is SQL Server 2008, and Matlab 2013a is used as algorithm simulation platform.The performance analysis and early warning system in mobile core network implemented in this paper, after on-site testing and adjustment in the branch company of China Mobiles in a certain city, has been on-site running for two years or so. The system obtains performance once every 15 minutes, and the time used each time is less than 2 minutes. Running results show that the various modules are working properly, and exception cases can be dealt automatically. The KPI index is calculated correctly, providing reliable basis for user to analyze network condition. The dynamic threshold algorithm proposed in this paper can accurately generate warning, providing the basis to find network failure for users, and is suitable for early warning of mobile network performance.