A method for smart grid intrusion detection through explainable deep learning

被引:0
|
作者
Ciaramella, Giovanni [1 ,2 ]
Martinelli, Fabio [3 ]
Santone, Antonella [4 ]
Mercaldo, Francesco [2 ,4 ]
机构
[1] IMT Sch Adv Studies Lucca, Lucca, Italy
[2] Natl Res Council Italy, Inst Informat & Telemat, Pisa, Italy
[3] Inst High Performance Comp & Networking, Arcavacata Di Rende, Italy
[4] Univ Molise, Campobasso, Italy
来源
JOURNAL OF COMPUTER VIROLOGY AND HACKING TECHNIQUES | 2025年 / 21卷 / 01期
关键词
Smart grid; Deep learning; Intrusion; Security;
D O I
10.1007/s11416-025-00549-1
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Over the years, cyber-attacks have increased drastically, and their execution changed with time. One of the targets of cyber criminals is trying to obtain sensitive information from mobile, cloud, or generally IoT devices. To avoid those risks, different countermeasures have been developed and implemented. For instance, the IEC 60870-5-104 protocol was developed to define the systems used for remote control in electrical engineering and power system automation applications. Starting from these considerations, in this paper, we propose an intrusion detector based on explainable Deep Learning (DL) that is able to detect possible attacks. In a nutshell, we consider several DL models, i.e., AlexNet, DenseNet, EfficientNet, Inception, LeNet, MobileNet, ResNet50, Standard CNN, VGG16, and VGG19 to understand whether a network trace (stored in a PCAP file) is related to an attack. Moreover, to explain of the model attack prediction, we resort to two different Class Activation Mapping algorithms available in the literature: Grad-CAM++ and Score-CAM. As the last step, we also calculated the IF/IM-SSIM index to strengthen the robustness of the top-performing model and evaluate the similarity between the two CAM algorithms. Experimental results show the effectiveness of the proposed method, and we obtained an accuracy equal to 0.900 with the DenseNet. In conclusion, we applied the exact steps to a new dataset to confirm that the proposed methodology is scalable and applicable to other datasets and achieved promising results.
引用
收藏
页数:20
相关论文
共 50 条
  • [21] FLDID: Federated Learning Enabled Deep Intrusion Detection in Smart Manufacturing Industries
    Verma, Priyanka
    Breslin, John G.
    O'Shea, Donna
    SENSORS, 2022, 22 (22)
  • [22] Proposed algorithm for smart grid DDoS detection based on deep learning
    Diaba, Sayawu Yakubu
    Elmusrati, Mohammed
    NEURAL NETWORKS, 2023, 159 : 175 - 184
  • [23] Enhanced and Explainable Deep Learning-Based Intrusion Detection in IoT Networks
    Gyawali, Sohan
    Sartipi, Kamran
    Van Ravesteyn, Benjamin
    Huang, Jiaqi
    Jiang, Yili
    MILCOM 2023 - 2023 IEEE MILITARY COMMUNICATIONS CONFERENCE, 2023,
  • [24] Ensemble Learning Methods for Anomaly Intrusion Detection System in Smart Grid
    Khoei, Tala Talaei
    Aissou, Ghilas
    Hu, When Chen
    Kaabouch, Naima
    2021 IEEE INTERNATIONAL CONFERENCE ON ELECTRO INFORMATION TECHNOLOGY (EIT), 2021, : 129 - 135
  • [25] Smart Grid Anomaly Detection using a Deep Learning Digital Twin
    Danilczyk, William
    Sun, Yan
    He, Haibo
    2020 52ND NORTH AMERICAN POWER SYMPOSIUM (NAPS), 2021,
  • [26] Intrusion Detection Systems in Smart Grid
    Rakas, Slavica Bostjancic
    Timcenko, Valentina
    Kabovic, Milenko
    Kabovic, Anka
    2022 21ST INTERNATIONAL SYMPOSIUM INFOTEH-JAHORINA (INFOTEH), 2022,
  • [27] An intelligent scheduling control method for smart grid based on deep learning
    Tong, Zhanying
    Zhou, Yingying
    Xu, Ke
    MATHEMATICAL BIOSCIENCES AND ENGINEERING, 2023, 20 (05) : 7679 - 7695
  • [28] Deep Learning in Intrusion Detection Systems
    Karatas, Gozde
    Demir, Onder
    Sahingoz, Ozgur Koray
    2018 INTERNATIONAL CONGRESS ON BIG DATA, DEEP LEARNING AND FIGHTING CYBER TERRORISM (IBIGDELFT), 2018, : 113 - 116
  • [29] Deep Learning in IoT Intrusion Detection
    Tsimenidis, Stefanos
    Lagkas, Thomas
    Rantos, Konstantinos
    JOURNAL OF NETWORK AND SYSTEMS MANAGEMENT, 2022, 30 (01)
  • [30] Explainable Ransomware Detection with Deep Learning Techniques
    Ciaramella, Giovanni
    Iadarola, Giacomo
    Martinelli, Fabio
    Mercaldo, Francesco
    Santone, Antonella
    JOURNAL OF COMPUTER VIROLOGY AND HACKING TECHNIQUES, 2024, 20 (02) : 317 - 330