基于自编码器的假冒攻击识别机制  

IMPERSONATION ATTACK IDENTIFICATION MECHANISM BASED ON AUTOENCODER

作　　者：杨建喜[1] 张媛利 蒋华[2] 朱晓辰 Yang Jianxi;Zhang Yuanli;Jiang Hua;Zhu Xiaochen(Beijing Electronic Science and Technology Institute,Beijing 100070,China;Xidian University,Xi’an 710071,Shaanxi,China)

机构地区：[1]北京电子科技学院,北京100070 [2]西安电子科技大学,陕西西安710071

出　　处：《计算机应用与软件》2022年第12期341-349,共9页Computer Applications and Software

基　　金：中央高校基本科研业务费项目(328201912)。

摘　　要：由于物联网(Internet of Things, IoT)中大量节点资源受限,导致传统的加密通信方式存在密钥分发时延、计算开销和系统开销大等问题。对此,提出一种基于自编码器(Autoencoder, AE)的假冒攻击识别机制。对不同用户根据不同信道特性建模,实现了当攻击用户冒充合法用户企图占用系统资源进行通信时,接收端利用K均值算法(K-means)对信号星座图及误码率(Bit Error Rate, BER)进行评估来及时监测并识别攻击用户,从而确保合法用户的正常通信。仿真分析表明,该方法有效地提高了攻击用户监测识别率,并且不存在传统通信方法中的加密时延等问题,适用于数据流量大的资源受限型节点处理器。Due to the limited resources of a large number of nodes in Internet of Things(IoT), there are many problems in traditional encrypted communication, such as key distribution delay, high computational overhead and system overhead. Therefore, an impersonation attack identification mechanism based on autoencoder(AE) was proposed. Different users were modeled with different channel characteristics. When attackers pretended to be legitimate users in an attempt to occupy system resources for communication, the receiver used K-means algorithm to evaluate the signal constellation and bit error rate(BER) to monitor and identify attackers timely, so as to ensure the normal communication of legitimate users. The simulation results show that this method can effectively improve the identification rate of attackers, and there is no encryption delay in traditional communication methods. It can be applied to resource-constrained node processors with huge data traffics.

关 键 词：IOT 假冒攻击 自编码器 K均值 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]