Robust Intrusion Detection and Recognition via Sparse Representation

• Published in: IET Intelligent Signal Processing Conference 2013 (ISP 2013) p. 7.P08
• Main Authors: Di Xu, Wei Li, Jie Zhu
• Format: Conference Proceeding
• Language: English
• Published: Stevenage, UK IET 2013
• Subjects: Optimisation techniques; Signal processing and detection; robust intrusion detection; interferometric optical fibre sensor; machine learning methods; antielectromagnetic interference; signal representation; signal classification; intrusion signals; optimisation; support vector machine; safety systems; optic fibre Sagnac interferometers; principal component analysis; sparse representation
• ISBN: 1849197741; 9781849197748
• DOI: 10.1049/cp.2013.2061

Due to its well-performed capabilities of anti-electromagnetic interference, low-cost and high-sensitivity, interferometric optical fibre sensor has been widely used in the security systems of oil and gas transportation these years. The main part among those systems is the classification of intrusion signals, obtained from optic fibre Sagnac interferometers. Based on sparse representation, we analyse the characteristics of the intrusion signals and argue that it offers the key to addressing this kind of classification problem. The most critical thing, however, is that whether we can acquire sufficient number of features and the right sparse representation. Combined with support vector machine(SVM), a new method is proposed for intrusion detection by casting the classification problem as finding sparse representation of testing samples with respect to training samples. To evaluate its performance, the proposed method is applied to lots of datasets and compared with several other machine learning methods.