Analysis of Classification Methods Based on Radio Frequency Fingerprint for Zigbee Devices

Radio frequency fingerprint is an inherent characteristic of wireless communication devices which can be extracted from communication signals and be applied in wireless device identification for communication system security. This paper selects different characteristics of RF fingerprints and compares the identification accuracy of Zigbee devices with five classification algorithms, including support vector machine, bagging, neural network, naive Bayes, and random forest algorithms. The experimental research shows that the highest identification accuracy reaches approximately 100% by using multi-features of frequency offset, IQ offset, and circle offset based on the neural network algorithm under high SNR. With the reduction in SNR, the identification accuracy based on bagging algorithm with multi-features of frequency offset and IQ offset is the highest. The performance of support vector machine algorithm is the most stable.