Communications Network Security Based on Radio Frequency Fingerprinting and Variational Mode Decomposition

Abstract

A technique that shows promise for physical layer security in wireless networks is radio frequency fingerprinting (RFF). RFF is a security solution for communication networks that relies on identifying the distinct characteristics of radio frequency transient signals. In this work, we assessed the RFF technique's performance using variational mode decomposition (VMD). The foundation of radio frequency fingerprinting (RFF) is the recognition of distinctive characteristics of RF transient signals that Bluetooth (BT) devices emit. BT devices' RF transient signals are non-stationary, nonlinear time series with brief durations. In order to do this, Bluetooth (BT) transient signals are broken down into a number of band-limited modes using VMD. The transient signal is then rebuilt from the modes. From the complicated form, higher order statistical (HOS) characteristics are retrieved of VMD-reconstructed transients. The BT devices are then identified using a classifier called the Linear Support Vector Machine (LVM). Experimental testing of the method has been conducted using BT devices from various mobile brand and model names. The classification performance for the same dataset shows that the (RFF with VMD) approach outperforms the (RFF without VMD) strategy by about 2.3%. Ultimately, it has been demonstrated that variational mode decomposition (VMD) provides an effective means of enhancing classification security and accuracy when utilized to extract features from Bluetooth (BT) transient signals.