摘要
为解决传统异常流量节点检测方法在复杂网络环境中误报率高等问题,设计一种基于多维特征融合与机器学习的混合检测方法。该方法针对流量突变型、协议违规型及行为异常型3类节点,构建了包含滑动窗口统计、深度包检测和马尔可夫链状态转移分析的三分支并行检测架构,实现了对异常节点的精确识别与分类。研究结果表明,该设计通过融合多维度特征,有效提升了异常流量节点检测的准确性与可靠性。
In order to solve the problem of high false alarm rate of traditional abnormal traffic node detection methods in complex network environment,a hybrid detection method based on multi-dimensional feature fusion and machine learning is designed.This method constructs a three-branch parallel detection architecture including sliding window statistics,deep packet detection and Markov chain state transition analysis for three types of nodes with sudden traffic change,protocol violation and abnormal behavior,and realizes accurate identification and classification of abnormal nodes.The research results show that the design effectively improves the accuracy and reliability of abnormal traffic node detection by integrating multi-dimensional features.
作者
姜宏伟
JIANG Hongwei(Ningcheng County Central Hospital,Chifeng City,Chifeng 024200,China)
出处
《通信电源技术》
2026年第2期142-144,共3页
Telecom Power Technology
关键词
计算机通信网络
异常流量节点
多维特征融合
混合检测模型
computer communication network
abnormal traffic node
multi-dimensional feature fusion
hybrid detection model
