Title: A network attack traffic identification method for power sensor network based on open-set recognition

Authors: Wei Liu; Qigang Zhu; Xingshen Wei; Junjiang He; Qiang Zhang; Tian Jiang; Zeji Sun

Addresses: School of Informatics, Xiamen University, Xiamen, China; NARI Group Corporation/State Grid Electric Power Research Institute, Nanjing, China ' NARI Group Corporation/State Grid Electric Power Research Institute, Guodian Nanrui Technology Co., Ltd., Building 1, Number 1, No. 19, Chengxin Avenue, Jiangning District, Nanjing, Jiangsu Province, China ' NARI Group Corporation/State Grid Electric Power Research Institute, Guodian Nanrui Technology Co., Ltd., Building 1, Number 1, No. 19, Chengxin Avenue, Jiangning District, Nanjing, Jiangsu Province, China ' School of Cyber Science and Engineering, Sichuan University, Chengdu, China ' School of Cyber Science and Engineering, Sichuan University, Chengdu, China ' NARI Group Corporation/State Grid Electric Power Research Institute, Guodian Nanrui Technology Co., Ltd., Building 1, Number 1, No. 19, Chengxin Avenue, Jiangning District, Nanjing, Jiangsu Province, China ' NARI Group Corporation/State Grid Electric Power Research Institute, Guodian Nanrui Technology Co., Ltd., Building 1, Number 1, No. 19, Chengxin Avenue, Jiangning District, Nanjing, Jiangsu Province, China

Abstract: With the expanding range of services offered by power sensor networks, precise identification of network traffic is indispensable for ensuring network security management and prevention. While machine learning-based and deep learning-based network traffic identification technology has advanced considerably, it remains constrained to classifying predetermined categories. In real-world network environments, novel attack types can surface that the trained model has not accounted for. These unforeseen factors can significantly degrade the performance of existing methods, making them inadequate to navigate the complexities of network environments. We introduce a network attack traffic identification model within power sensor systems to address the above challenges, leveraging the autoencoder. By harnessing the intrinsic properties of the autoencoder alongside tailored thresholds, the model effectively accomplishes open-set recognition of network traffic. The experimental results highlight the model's strong performance in open-set recognition scenarios and confirm its effectiveness in power sensor applications.

Keywords: network attack traffic detection; open-set recognition; OSR; power sensor network.

DOI: 10.1504/IJSNET.2025.144632

International Journal of Sensor Networks, 2025 Vol.47 No.3, pp.173 - 185

Received: 14 Sep 2024
Accepted: 30 Sep 2024
Published online: 25 Feb 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article