基于图神经网络与量子密钥分发协议的局域网安全通信技术

施金诺1; 熊贵云1; 童江艳2

基于图神经网络与量子密钥分发协议的局域网安全通信技术

引用本文：施金诺1，熊贵云1，童江艳2.基于图神经网络与量子密钥分发协议的局域网安全通信技术[J].计算技术与自动化,2025,(3):178-182

摘要点击次数: 327

全文下载次数: 229

作者	单位
施金诺1，熊贵云1，童江艳2	（1.云南省中共怒江州委网信办,云南怒江 673100；2.云南省中共怒江州委宣传部,云南怒江 673100）

中文摘要:当前局域网安全通信过程中，主要通过密钥分发中心（KDC）对密钥进行集中管理与分发，其分发模式是固定不变的，导致遇到攻击后局域网通信丢包率较高。因此，提出了基于图神经网络与量子密钥分发协议的局域网安全通信技术。运用椭圆曲线密码算法处理网络通信数据，得到一系列加密会话。针对加密流量进行过滤处理后，考虑TLS握手原始字节和TLS记录长度序列，生成单条加密会话的特征表示向量。构建基于图神经网络的智能分类模型，将加密流量特征表示输入其中，即可检测出恶意流量，将其从加密会话中剔除。最后，引入量子密钥分发协议帮助通信双方生成共享密钥，安全解析加密会话，实现局域网安全通信。实验结果表明：该技术实施后，局域网安全通信过程中产生的丢包率不超过0.2%，满足了网络信息传输要求。

中文关键词:局域网加密图神经网络恶意流量量子密钥分发协议安全通信

Lan Secure Communication Technology Based on Graph Neural Network and Quantum Key Distribution Protocol

Abstract:In the current process of secure communication in local area networks, keys are mainly centrally managed and distributed through the key distribution center(KDC), and its distribution mode is fixed and unchanged, resulting in a high packet loss rate in local area network communication after encountering attacks. Therefore, a secure communication technology for local area networks based on graph neural networks and quantum key distribution protocols is proposed. Using elliptic curve cryptography algorithm to process network communication data and obtain a series of encrypted sessions. After filtering encrypted traffic, consider the TLS handshake raw bytes and TLS record length sequence to generate a feature representation vector for a single encrypted session. Constructing an intelligent classification model based on graph neural networks, inputting encrypted traffic feature representations into it, can detect malicious traffic and remove it from encrypted sessions. Finally, the introduction of quantum key distribution protocol helps communication parties generate shared keys, securely parse encrypted sessions, and achieve secure communication in local area networks. The experimental results show that after the implementation of this technology, the packet loss rate generated during secure communication in the local area network does not exceed 0.2%, meeting the requirements of network information transmission.

keywords:local area network encryption graph neural network malicious traffic quantum key distribution protocol secure communication

查看全文 查看/发表评论 下载pdf阅读器