The Effect of Human Body Shadowing in ZigBee Radio Frequency Fingerprinting Identification

Abstract

Security is a major concern for the Internet of Medical Things (IoMT), however, many of these devices are limited in their capabilities. Physical layer (PHY) security measures can be used to prevent unauthorized access by exploiting intrinsic emitter characteristics such as Specific Emitter Identification (SEI), known as Radio Frequency Fingerprinting Identification (RFFI). RFFI at the IoMT is a promising approach to secure wearable Body Sensor Network (WBSN) devices. In this paper, we evaluated the effect of human body shadowing on Radio Frequency Fingerprinting Identification (RFFI) systems. Results show that shadowing has a serious impact on RFFI models. We also show that it can be mitigated by applying log-normal shadowing augmentation. Results obtained from simulations and experimental trials show that the classification accuracy increases when the multipath channel model and shadowing block size of 640 are used. A new system model for classifying devices using RFFI is then proposed. The proposed model achieved better classification accuracy when evaluated using unseen shadowed data.