Identification Mining of Unusual Patterns for Multimedia Communication Networks by Using Fuzzy Linear Programming

摘要

This paper presents a data mining system of performance evaluation for multimedia communication networks (MCNs). The architecture of the data mining system for performance evaluation and the snowflake schema of data warehouse for performance evaluation are built. Then this paper applies the fuzzy classification methods to identify that the MCNs are in unusual patterns. Two important performance evaluation problems for the MCNs are considered in this paper. They are: (1) optimization problem for construction of the data mining system of performance evaluation; (2) category with real-time data corresponding the data mining system with means of dividing the performance data into usual and unusual categories. An algorithm is employed to identify performance data such as throughput capacity, package forwarding rate, response time and so on. A software named PEDM2.0 (Performance Evaluation Data Miner) is proposed to improve the accuracy and the effectiveness of the fuzzy linear programming (FLP) method comparing with decision tree, neural network, multiple criteria linear programming.