基于循环神经网络的光通信网络节点入侵告警算法

A Node Intrusion Warning Algorithm for Optical Communication Networks Based on Recurrent Neural Networks

当前光通信网络节点在遭受入侵时,告警的准确性和响应速度都存在较大的不足,为此,开展基于循环神经网络的光通信网络节点入侵告警算法研究。收集光通信网络的数据,从中提取受损节点的特征。同时,利用循环神经网络中的长短期记忆网络(Long Short-Term Memory,LSTM)模型,并利用Sigmoid函数补偿信息,构建改进LSTM的节点入侵检测模型,完成入侵行为告警。实验结果表明,设计方法可以实现光通信网络节点入侵行为的准确告警,且响应速度明显提高。

When nodes in current optical communication networks are invaded,there are significant shortcomings in the accuracy and response speed of alarms.Therefore,the algorithm of node intrusion alarm of optical communication network based on recurrent neural network is studied.Collect data from optical communication networks and extract features of damaged nodes from them.At the same time,utilizing the Long Short Term Memory(LSTM)model in recurrent neural networks and utilizing the Sigmoid function to compensate for information,an improved LSTM node intrusion detection model is constructed to complete intrusion behavior alerts.The experimental results show that the design method can achieve accurate alarm of node intrusion behavior in optical communication networks,and the response speed is significantly improved.