Adaptive Physical-Layer Authentication for IoT in MIMO Communication Systems Using Support Vector Machine
Mohammed Abdrabou, T. Aaron Gulliver
Abstract
Internet of Things (IoT) devices have become ubiquitous due to the prevalence of the Internet. However, the open nature of wireless networks makes them susceptible to spoofing attacks. Moreover, their heterogeneous characteristics create authentication challenges which are increasing due to the tremendous growth in the number and variety of devices. Physical-layer authentication (PLA) provides a solution by utilizing the unique characteristics of wireless channels to aid upper layer authentication (ULA). In this article, an adaptive PLA scheme is proposed which exploits the antenna diversity inherent in multiple-input multiple-output (MIMO) systems. This scheme employs a one-class classification support vector machine (OCC-SVM) with the magnitude and real and imaginary parts of the received signals as features. Results are presented which show that this scheme provides robust authentication. The authentication performance is evaluated considering two majority voting schemes for IoT applications.