Backdoor Attacks Prediction in IIoT Network using Optimal Double Mask Region Convolution Model

Nalini SubramanianS. Nikkath BushraG. ShobanaS. Radhika1 Rajalakshmi Engineering College,Thandalam,India2 Department of IT,St. Joseph's Institute of Technology,Chennai,India3 Department of Computer Applications,Madras Christian College,Tambaram,Chennai,India4 School of Electrical and Electronics Engineering,Sathyabama Institute of Science and Technology,Chennai,119,IndiaNalini Subramanian received her PhD from Sathyabama Institute of Science and Technology in 2020. She obtained her ME degree in computer science & engineering from Sathyabama University,Chennai,India in 2006. She has 17 years of teaching experience. She is currentlyworking in Rajalakshmi Engineering College,India. Her area of research interest includes cloud computing,network security and machine learning.S Nikkath Bushra received her PhD from Bharath Institute of Higher Education and Research in 2017. She obtained her ME degree in computer science & engineering fromSathyabamaUniversity,Chennai,India in 2006. She has 17 years of teaching experience. She is currently working in St. Joseph's Institute of Technology and has published more than 20 research articles. Her area of research interest includes cloud computing,big data analytics,machine learning and deep learning. Email: Shobana is currently working as assistant professor in the Department of Computer Applications,Madras Christian College,Chennai. She completed her ME (computer science and engineering) in 2006 from Sathyabama University,Chennai. She completed her masters in bioinformatics from Bharathiar University with distinction. She has more than 17 years of teaching experience and has publishedmore than 20 research articles. Her research areas include machine learning,bioinformatics,cheminformatics and cloud computing. Email: gmshobana@gmail.comS Radhika is a associate professor from School of Electrical and Electronics Engineering of Sathyabama Institute of Science and Technology since 2006. She completed her PhD with research title "Design of Adaptive Filtering Algorithms for Acoustic Echo Cancellation Application". Her areas of research include adaptive signal processing,system identification,echo cancelation and sparse signal processing. She has published several articles in international and national journals and conferences related to the adaptive filter algorithms. Email: radhikachandru79@gmail.com

DOI: https://doi.org/10.1080/03772063.2023.2230174

IF: 1.8768

2024-02-08

IETE Journal of Research

Abstract:The evolution of Industrial Internet of Things (IIoT) applications provides intelligent microservices to process the generated massive data. However, the open and interconnected structure of IIoT makes the network model more vulnerable to malware attacks. Among different kinds of threats, the backdoor attack is considered the main imperceptible attack due to its unobservable and opaque characteristics. Therefore, to accurately predict backdoor attacks in the network model and to mitigate them efficiently, this paper proposes a novel concept named "novel binary knowledge gaining based double mask region convolution (NBG-DRC) model". The main objective of the proposed concept is two-fold: backdoor attack prediction and trigger identification. The backdoor attack prediction module uses a double mask region convolution (DRC) network to predict whether the data is malicious or normal. Subsequently, in the trigger identification phase, the structure and positions of backdoor triggers are identified accurately using the NBG-DRC model. The proposed model uses four different datasets namely CIFAR-10, MNIST, CIFAR-100, and GTSRB for analysis. The efficiency of the proposed NBG-DRC model is examined by comparing its performance rate with other compared techniques in terms of evaluation metrics namely false negative rate, false positive rate, computational overhead, execution time, accuracy, and success rate. The proposed NBG-DRC technique achieves a greater accuracy percentage of about 94% for the CIFAR-10 dataset, 98.7% for the MNIST dataset, 91% for the CIFAR-100 dataset, and 97.23% for the GTSRB dataset.

telecommunications,engineering, electrical & electronic