IPV6多跳网络环境下双通道快速切换算法的构建研究  

Research on the construction of dual-channel fast switching algorithm in IPV6 multi-hop network environment

作　　者：王祝先[1] 赵忠凯 叶润泽 关兴民[1] 杨智涛[1] 宋邦钰 WANG Zhuxian;ZHAO Zhongkai;YE Runze;GUAN Xingmin;YANG Zhitao;SONG Bangyu(Heilongjiang Provincial Meteorological Data Centre,Harbin 150030,China)

机构地区：[1]黑龙江省气象数据中心,黑龙江哈尔滨150030

出　　处：《应用科技》2024年第5期101-106,共6页Applied Science and Technology

摘　　要：多跳网络环境下,节点的复杂性通常会对网络技术的适配性带来极大的挑战。为了提高其适应性,提出了一种双通道快速切换算法。该算法采用双通道设计和半监督学习策略,结合多头注意力机制,提高节点分类的精度和效率,同时优化网络的响应和切换时间。实验结果证明,该方法的节点分类准确率达到95.56%,相较于卷积神经网络提高35.56%~62.23%;其最佳响应时间为0.22 s,相较于卷积神经网络领先了0.52~1.03 s;最佳切换时间为0.89 s。该方法的提出和实施,尤其是在节点分类精度和网络响应时间方面的优异表现,为多跳网络IPv6技术的适应性研究提供参考思路。In the multi-hop network environment,the complexity of nodes usually brings great challenge to the adaptation of network technology.In order to improve its adaptability,a two-channel fast switching algorithm was proposed.The two-channel design and semi-supervised learning strategy,combined with multi-head attention mechanism,are designed to improve the accuracy and efficiency of node classification,and optimize the response and switching time of network.The experimental results show that the node classification accuracy of this method reaches 95.56%,which is 35.56%~62.23%higher than that of convolutional neural networks;the optimal response time is 0.22 s,which is 0.52~1.03 s higher than that of convolutional neural network;the optimal switching time is 0.89s.The proposal and implementation of this method,especially its excellent performance in node classification accuracy and network response time,provide a reference for the adaptability research of multi hop network IPv6 technology.

关 键 词：快速切换 双通道 注意力机制 半监督学习 IPV6 图神经网络 节点分类 算法构建 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置]