Multiscaleanalysis of Skewness for Feature Extraction Inreal-Time

摘要

This paper describes a generalized multiscale analysis methodology with applications in cybersecurity. This research looks for finding applicability of multiscale analysis in real-time feature extraction. The generalized multiscale analysis methodology introduced here can utilize an optimal mathematical operator for searching features within a signal. The practical application of the generalized multiscale methodology in this research is shown addressing the third higher order moment, skewness. Monoscale analysis follows the conventional treatment of sequences connected with most of the signal processing being done in the traditional monoscale ecosystem. Hence, monoscale analysis utilizes all the information available within an epoch, which when acquired satisfies the Nyquist sampling frequency. In the last decades, fresh and untraditional views have refined fractal approaches for measurements and the conception of the multiscale analysis in signal processing has been proposed by this research group and used extensively. Multiscale analysis is required in cybersecurity because it allows searching for information, which may be scattered at different scales, in order to fingerprint anomalous activity in Internet/network traffic.