基于多模态特征的新型电力系统终端无线指纹提取与识别

Novel Wireless Fingerprint Extraction and Identification of Power System Terminal Based on Multimodal Features

随着新型电力系统的不断发展,设备在开放环境中存在被假冒攻击的可能。常见的应对方式是基于无线信号指纹的识别技术,在物理层面保证设备的安全性。然而,由于无线信号的不稳定性,导致对终端设备的识别准确率较低。针对上述问题,文章提出了一种基于多模态特征的新型电力系统终端无线指纹提取与识别方法。通过融合无线信号的时域特征和频域特征构建动态指纹数据库,形成一种基于多模态特征的新型电力系统无线终端设备识别认证的方法,实现了稳定的无线终端指纹提取,可以有效在无线设备接入时和无线设备持续通信期间对无线设备进行识别认证,保证无线设备的安全可靠运行。通过不同场景下的实验结果表明,使用帧响应时间间隔作为时域特征和信道状态信息作为频域特征的多域融合方法进行无线设备识别,可以显著提高计算的准确性和稳定性,识别率达到95%以上。

With the continuous development of a new type of power system,there is a possibility of counterfeit attacks on devices in open environments.A common response is the identification technology based on wireless signal fingerprinting,which ensures the security of the devices at the physical level.However,the instability of wireless signals leads to a low accuracy of identification of terminal devices.Aiming at the above problems,thia paper proposes a novel wireless fingerprint extraction and recognition method for power system terminals based on multimodal features.A dynamic fingerprint database is constructed by fusing the time-domain features and frequency-domain features of wireless signals to form a new type of power system wireless terminal device identification and authentication method based on multimodal features,which realizes stable wireless terminal fingerprint extraction,and can effectively identify and authenticate the wireless devices during wireless device access and wireless device continuous communication to ensure the safe and reliable operation of wireless devices.The experimental results in different scenarios show that the multi-domain fusion method using the frame response time interval as a time-domain feature and the channel state information as a frequency-domain feature for the identification of wireless devices can significantly improve the accuracy and stability of the computation,and the identification rate reaches more than 95%.