Research And Implementation Of The System For Telecommunication Traffic Monitoring, Analysis And Mining

The runtime information of telecommunication traffic is crucial important in running and maintenance of telecommunication network because it provides decision aiding of information for hardware construction, equipement updating, breakdown maintenance, traffic schedule and resource assignment. The most importance of the runtime information of telecommunication traffic is that the customer Expansion and traffic incomes can be got through analysis. So, to monitor, analyse and mining telecommunication traffic can help managers make right decisions, guide expansion and optimization of network, guarantee the safe running and maintenance of telecommunication network. It has great social and economical significance.This study aims at the shortcoming of traditional monitoring and analysing system for telecommunication traffic and impliment a monitoring, analysing and mining system for telecommunication traffic. We use machine learning methods, which provide an efficient pattern for running telecommunication network in intelligent ways, to predict telecommunication traffic and classify machine troubles intelligently. We focus on the functional implements of the system and relative key technology and methods to study and discuss. The main content we studied includes: the software engineering methods in the process of system construction, the intelligent predicting methods for telecommunication traffic based on k-nearest neighbour methods and the machine trouble classifying technology based on naive Bayesian methods.The system uses moducle designing methods and is divided into 5 subsystem including user permission administrator, trouble classifying and handling, rules and regulations, telecommunication traffic analysis and mining and electronic resources querying. We choose object-oriented and extendable coding language environment of ASP.NET2.0 and Sq1 Server 2005 database to implement that functions.This study proposes an intelligent predicting method for telecommunication traffic based on k-nearest neighbor algorithm, which transform the telecommunication traffic data into input instances and adopt a distance metric based on feature weighting. This method can predict telecommunication traffic efficiently and precisely.This study also raises a trouble classifying technology based on naive Bayesian methods. It decomposes the trouble description into multi-features using vector space model and calculates the probability of all possible trouble types using Bayesian method. This method can estimate trouble type precisely and provide the probability of all trouble types.