Dynamic IoT Device Identification Through Single-Packet Feature Analysis

With the rapid development of the Internet of Things (IoT), concerns about the security of IoT devices are becoming increasingly evident. To achieve timely detection and isolation of suspicious devices and thereby protect IoT devices, device identification has emerged. However, existing research still faces challenges such as the transfer problem and poor dynamic adaptability. In this study, we propose a novel device identification approach based on single-packet features, named SPDI (Single-Packet Device Identification). It can achieve efficient and accurate device identification through two-stage classification. In the firststage, a multi-classifier based on the One-vs-All (OvA) strategy is trained using the optimal feature subset. If a tie occurs in the firststage, the process proceeds to the second-stage, where the final decision is made by calculating the edit distance of the Local Sensitive Hashing (LSH) values of the payload. Experimental results show that this method performs excellently on Aalto and UNSW datasets, with accuracy and F1 scores surpassing those of existing methods.