Ownership protection of outsourced biomedical time series data based on optimal watermarking scheme in data mining

Trung Duy PHAM, Dat TRAN, Wanli MA

    Research output: Contribution to journalArticle

    2 Citations (Scopus)
    8 Downloads (Pure)

    Abstract

    In the biomedical and healthcare fields, the ownership protection of the outsourced data is becoming a challenging issue in sharing the data between data owners and data mining experts to extract hidden knowledge and patterns. Watermarking has been proved as a right-protection mechanism that provides detectable evidence for the legal ownership of a shared dataset, without compromising its usability under a wide range of data mining for digital data in different formats such as audio, video, image, relational database, text and software. Time series biomedical data such as Electroencephalography (EEG) or Electrocardiography (ECG) is valuable and costly in healthcare, which need to have owner protection when sharing or transmission in data mining application. However, this issue related to kind of data has only been investigated in little previous research as its characteristics and requirements. This paper proposes an optimized watermarking scheme to protect ownership for biomedical and healthcare systems in data mining. To achieve the highest possible robustness without losing watermark transparency, Particle Swarm Optimization (PSO) technique is used to optimize quantization steps to find a suitable one. Experimental results on EEG data show that the proposed scheme provides good imperceptibility and more robust against various signal processing techniques and common attacks such as noise addition, low-pass filtering, and re-sampling.
    Original languageEnglish
    Article number1541
    Number of pages25
    JournalAustralasian Journal of Information Systems
    Volume21
    DOIs
    Publication statusPublished - 2017

    Fingerprint

    Watermarking
    Data mining
    Time series
    Electroencephalography
    Electrocardiography
    Transparency
    Particle swarm optimization (PSO)
    Signal processing
    Sampling
    Time series data
    Ownership
    Healthcare
    Owners

    Cite this

    @article{00a77a9730b44e6e85b8b10b2e36e98a,
    title = "Ownership protection of outsourced biomedical time series data based on optimal watermarking scheme in data mining",
    abstract = "In the biomedical and healthcare fields, the ownership protection of the outsourced data is becoming a challenging issue in sharing the data between data owners and data mining experts to extract hidden knowledge and patterns. Watermarking has been proved as a right-protection mechanism that provides detectable evidence for the legal ownership of a shared dataset, without compromising its usability under a wide range of data mining for digital data in different formats such as audio, video, image, relational database, text and software. Time series biomedical data such as Electroencephalography (EEG) or Electrocardiography (ECG) is valuable and costly in healthcare, which need to have owner protection when sharing or transmission in data mining application. However, this issue related to kind of data has only been investigated in little previous research as its characteristics and requirements. This paper proposes an optimized watermarking scheme to protect ownership for biomedical and healthcare systems in data mining. To achieve the highest possible robustness without losing watermark transparency, Particle Swarm Optimization (PSO) technique is used to optimize quantization steps to find a suitable one. Experimental results on EEG data show that the proposed scheme provides good imperceptibility and more robust against various signal processing techniques and common attacks such as noise addition, low-pass filtering, and re-sampling.",
    keywords = "time series, watermarking, Particle Swarm Optimization, EEG",
    author = "PHAM, {Trung Duy} and Dat TRAN and Wanli MA",
    year = "2017",
    doi = "10.3127/ajis.v21i0.1541",
    language = "English",
    volume = "21",
    journal = "Australian Journal of Information Systems",
    issn = "1039-7841",
    publisher = "UQ Business School, The University of Queensland",

    }

    TY - JOUR

    T1 - Ownership protection of outsourced biomedical time series data based on optimal watermarking scheme in data mining

    AU - PHAM, Trung Duy

    AU - TRAN, Dat

    AU - MA, Wanli

    PY - 2017

    Y1 - 2017

    N2 - In the biomedical and healthcare fields, the ownership protection of the outsourced data is becoming a challenging issue in sharing the data between data owners and data mining experts to extract hidden knowledge and patterns. Watermarking has been proved as a right-protection mechanism that provides detectable evidence for the legal ownership of a shared dataset, without compromising its usability under a wide range of data mining for digital data in different formats such as audio, video, image, relational database, text and software. Time series biomedical data such as Electroencephalography (EEG) or Electrocardiography (ECG) is valuable and costly in healthcare, which need to have owner protection when sharing or transmission in data mining application. However, this issue related to kind of data has only been investigated in little previous research as its characteristics and requirements. This paper proposes an optimized watermarking scheme to protect ownership for biomedical and healthcare systems in data mining. To achieve the highest possible robustness without losing watermark transparency, Particle Swarm Optimization (PSO) technique is used to optimize quantization steps to find a suitable one. Experimental results on EEG data show that the proposed scheme provides good imperceptibility and more robust against various signal processing techniques and common attacks such as noise addition, low-pass filtering, and re-sampling.

    AB - In the biomedical and healthcare fields, the ownership protection of the outsourced data is becoming a challenging issue in sharing the data between data owners and data mining experts to extract hidden knowledge and patterns. Watermarking has been proved as a right-protection mechanism that provides detectable evidence for the legal ownership of a shared dataset, without compromising its usability under a wide range of data mining for digital data in different formats such as audio, video, image, relational database, text and software. Time series biomedical data such as Electroencephalography (EEG) or Electrocardiography (ECG) is valuable and costly in healthcare, which need to have owner protection when sharing or transmission in data mining application. However, this issue related to kind of data has only been investigated in little previous research as its characteristics and requirements. This paper proposes an optimized watermarking scheme to protect ownership for biomedical and healthcare systems in data mining. To achieve the highest possible robustness without losing watermark transparency, Particle Swarm Optimization (PSO) technique is used to optimize quantization steps to find a suitable one. Experimental results on EEG data show that the proposed scheme provides good imperceptibility and more robust against various signal processing techniques and common attacks such as noise addition, low-pass filtering, and re-sampling.

    KW - time series

    KW - watermarking

    KW - Particle Swarm Optimization

    KW - EEG

    UR - http://journal.acs.org.au/index.php/ajis/article/view/1541

    U2 - 10.3127/ajis.v21i0.1541

    DO - 10.3127/ajis.v21i0.1541

    M3 - Article

    VL - 21

    JO - Australian Journal of Information Systems

    JF - Australian Journal of Information Systems

    SN - 1039-7841

    M1 - 1541

    ER -