The Research Of A Mobile Communication Charging Risk System

How to effectively control the risk of arrears, is one of most hot topics of the domestic and foreign operators of mobile communication industry.Tibet Mobile using the quasi-real-time billing, and in fact this is a pattern of off-line billing, which means it need of collecting off-line dialog list, and then analyzed, approved prices, storage, handling a series of accounts to achieve the purpose of billing. This model does not have the real-time billing control mechanisms, there is the possibility of overdraft users, easy to arrears problems, leading to the loss of income.For the arrears problem, the current operator of the measures adopted are: the adoption of intelligent gateways to control the user's call arrears, but need write user's frequent changes data into intelligent network hardware, this is a very big bottleneck maintenance; Grant price and account after the downtime has been done to deal with the arrears, this method can prevent users from continuing arrearage but can't to predict user's Arrearage; online charging (ocs), for all users that occurs during call arrears, indiscriminate cut off users'call, easy to hurt the integrity clients'and Major clients'feelings, resulting in the loss of customers.How to effectively control arrearage and prevent the loss of customers, it will be the focus of this paper. In this paper, use data mining techniques to effective classify mobile phone users, and in accordance with the classification results of the user to set the amount of overdraft of the arrears. Collecte data of part of users'non-payment history ,and use the decision tree data mining algorithm to get arrearage model, and use this model to classify all users. In this paper, I use CART algorithm, C4.5 algorithm and ID3 algorithm in telecommunications comparison of static data to identify the most suitable algorithm for static data, and then combine this algorithm with the incremental learning algorithm to improve this algorithms to deal with incremental dynamic data to predict the risk of arrears.Directly or indirectly become overdrawn consumption threshold, and timing of the output into a oracle database table.In this paper, learn Tibet Mobile BOSS (Business Operation Support System) status quo, studied the realization to improve this system, and design a set of background procedures for the realization of charging risk control system: studied the algorithm of the load balancing in IN-BOSS module, and achieved load balancing scheduling process; in anti-counting procedures to calculate the above-mentioned data mining algorithm received threshold table, and so on. By the test situation, this system can real-time cut off the low credit arrears when his calling, and may be allowed the users with high-credit degree in arrears to continue call a certain amount. In this paper, however, this improved CART algorithm to predict the outcome is dependent on the input parameters, how to optimize the choice of input parameters is the need to continue study.