Fuzzy Based Diagnostics System for Identifying Network Traffic Flow Anomalies

In recent years, much work has been constructed in the area of tool development in order to ease a network administrator’s job. However, there lack tools to collect and process flow data efficiently. This paper discusses the usage of network traffic properties in passive network monitoring which are used in recognizing and identifying anomaly. A fuzzy based diagnostic system imbedded with properties to recognize and identify network operation anomaly intelligently along with Neural Network as tuner has been proposed in this paper. This paper focuses on constructing a fuzzy system by manipulating the decoded data packets (inputs) to identify anomalies. Aspects such as the selection of a suitable fuzzy set operation and tuning it have proved to increase the reliability of the computed result. In this approach, Takagi Sugeno’s Fuzzy model has been implemented. With this fuzzy model, network operation anomalies are detected in accord with the intensity of the anomaly. This model also has the capability of choosing the suitable type of alerts; log, email or sms. By incorporating the fuzzy model with neural network, network operators are able to spend more time troubleshooting faults, thus minimizing the downtime of a particular segment in a network.