e-space
Manchester Metropolitan University's Research Repository

    IADF-CPS: Intelligent Anomaly Detection Framework towards Cyber Physical Systems

    Nagarajan, Senthil Murugan, Deverajan, Ganesh Gopal, Bashir, Ali Kashif ORCID logoORCID: https://orcid.org/0000-0001-7595-2522, Mahapatra, Rajendra Prasad and Al-Numay, Mohammed S (2022) IADF-CPS: Intelligent Anomaly Detection Framework towards Cyber Physical Systems. Computer Communications, 188. pp. 81-89. ISSN 0140-3664

    [img]
    Preview
    Accepted Version
    Available under License Creative Commons Attribution Non-commercial No Derivatives.

    Download (2MB) | Preview

    Abstract

    Cyber–Physical Systems (CPSs) becoming one of the most complex, intelligent, and sophisticated system. Ensuring security is an important aspect towards CPSs. However, increase in sophisticated and complexity attacks in CPSs, the conventional anomaly detection methods are facing problems and also growth in volume of data becomes challenging which requires domain specific knowledge that could be applied directly to analyze these challenges. In order to overcome this problem, various deep learning based anomaly detection system is developed. In this research, we propose an anomaly detection approach by integration of intelligent deep learning technique named Convolutional Neural Network (CNN) with Kalman Filter (KF) based Gaussian-Mixture Model (GMM). The proposed model is used for identifying and detecting anomalous behavior in CPSs. This proposed framework consists of two important process. First is to pre-process the data by transforming and filtering original data into new format and achieved privacy preservation of the data. Secondly, we proposed GMM-KF integrated deep CNN model for anomaly detection and accurately estimated the posterior probabilities of anomalous and legitimate events in CPSs.

    Impact and Reach

    Statistics

    Activity Overview
    6 month trend
    114Downloads
    6 month trend
    48Hits

    Additional statistics for this dataset are available via IRStats2.

    Altmetric

    Actions (login required)

    View Item View Item