基于分形理论的光纤网络流量异常检测技术

Anomaly Detection Technology of Fiber Network Traffic a Based on Fractal Theory

异常检测是光纤通信中需要解决的核心问题,为了准确检测出光纤通信中的网络流量异常行为,保证光纤网络的正常运行,提出一种基于分形理论的光纤网络流量异常检测技术。首先采用分形理论对光纤网络流量进行分析,获得网络的延迟时间向量,然后正交基神经网络建立网络流量异常检测模型,并将模型参数看作一种函数优化问题,最后采用教学优化算法对参数优化问题求解。实验结果表明,本文方法不仅提高了光纤网络流量异常检测的精度,而且对一些异常行为检测效果要明显优于当前其它光纤网络异常检测方法。

Anomaly detection is the core problem need to solve the demand for optical fiber communication,in order to detect network traffic anomaly behavior in optical fiber communication to ensure the normal operation of the optical fiber network,the optical fiber network traffic anomaly inspection technology based on fractal theory is proposed in this paper.Firstly,the fractal theory is used to analyze the optical fiber network traffic,network delay time vector is obtained and use orthogonal basis neural network to establish network traffic anomaly detection model,and the model parameters as a function optimization problem,and finally adopt the teaching optimization algorithm for parameter optimization problem solving.The experimental results show that this method improves the accuracy of the fiber network traffic anomaly detection,and it is better than the other optical fiber network anomaly detection methods.