| 引用本文: | 叶欣茹,伍益明,徐明,郑宁.基于深度学习的智能交通系统通信网络脆弱性检测[J].控制理论与应用,2022,39(10):1872~1880.[点击复制] |
| YE Xin-ru,WU Yi-ming,XU Ming,ZHENG Ning.Communication network vulnerability detection of intelligent transportation systems based on deep learning[J].Control Theory and Technology,2022,39(10):1872~1880.[点击复制] |
|
| 基于深度学习的智能交通系统通信网络脆弱性检测 |
| Communication network vulnerability detection of intelligent transportation systems based on deep learning |
| 摘要点击 1092 全文点击 459 投稿时间：2021-09-30 修订日期：2022-10-30 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2022.10930 |
| 2022,39(10):1872-1880 |
| 中文关键词 智能交通系统 网络安全 脆弱性检测 深度学习 |
| 英文关键词 intelligent transportation systems network security vulnerability detection deep learning |
| 基金项目 国家自然科学基金,浙江省公益技术应用研究项目 |
|
| 中文摘要 |
| 智能交通系统是集群智能技术的典型应用之一. 为解决现有智能交通通信网络脆弱性检测方法复杂度高、实时性差的问题, 提出引入深度学习技术对网络脆弱性检测方法进行设计. 先利用多智能体网络协同和消息传输机制与智能交通系统车辆间协作通信网络的共通性, 将智能交通系统通信图脆弱性检测问题建模为对多智能体网络r-鲁棒值的求解问题. 再针对随网络节点数目增多r-鲁棒值求解成NP难问题, 设计给出一种融入残差网络的深度学习算法, 将鲁棒值求解问题转化为深度学习图分类问题. 所提算法可有效应对动态多变的智能交通通信网络并对其实现快速精准的脆弱性检测. 最后通过一组典型交通场景的仿真实验验证本文所提方法的有效性. |
| 英文摘要 |
| Intelligent transportation system is one of the typical applications of swarm intelligence technology. In order to solve the problems of high complexity and poor real-time performance of existing vulnerability detection methods, a deep learning approach for vulnerability detection in intelligent transportation network is proposed. Firstly, based on the commonality of multi-agent network cooperation mechanism and intelligent transportation system cooperative network, the network vulnerability detection problem of intelligent transportation system is modeled as the problem of solving the r-robust value of multi-agent network. Then, as solving the r-robust value problem is NP-Complete, a deep learning algorithm integrated with residual network is designed, and the robust value solving problem is transformed into a deep learning graph classification problem. The proposed algorithm can effectively deal with the dynamic intelligent transportation communication network and realize fast and accurate vulnerability detection for the network. Finally, the effectiveness of the proposed method is verified by a set of simulation experiments in typical traffic scenes. |
|
|
|
|
|