Research And Implementation Of IoT Wireless Device Fingerprint Identification Technology

With the rapid development of the Internet of Things,wireless technology has gradually become an indispensable part of modern computing platforms and embedded systems.In order to improve the security of device access authentication and communication process in the Io T scenario,as well as the availability of ancillary services such as positioning,asset tracking,and travel time estimation based on wireless protocols,wireless device fingerprint identification is considered a promising solution.However,the extraction of the features from network layer and its upper layers often requires that the device to be identified is associated with a certain wireless network and can obtain the plaintext of the payload.However,BLE and WiFi packets have been encrypted above the data link layer,which makes those features difficult to extract.In response to these problems,this thesis is dedicated to researching the identification of BLE and WiFi devices based on the fingerprint features of the data link layer and the physical layer.The main work is as follows:Firstly,a BLE device category identification technology based on the fingerprint of the link layer broadcast packet is proposed.The thesis uses software-defined radio to collect BLE broadcast packets,extract fields that reflect device differences through traffic analysis.After excluding explicit identifiers,such as MAC addresses and device name,the multilayer perceptron model is used to identify BLE devices’ s category,which is defined as a combination of type and manufacturer.The identification accuracy of 15 types of BLE devices can reach99.8%.Secondly,the WiFi device identification technology based on the fingerprint of link layer probe request is researched.The fields reflecting the differences in hardware and software configuration of the chip in the probe request are analyzed,and the IEs and duration fields are extracted,then the existence of various IEs is coding as features.After removing the easily modified or unstable fields such as MAC address,SSID and DSSS parameter set,15 WiFi devices are preliminarily identified by using the multilayer perceptron model,and the identification accuracy of 11 devices reaches 100%.Thirdly,in order to realize the individual identification of WiFi devices,the WiFi device identification technology based on multi-domain physical layer fingerprints is studied.Waveform domain features based on LTS and STS of frame preamble are proposed.Statistical features and MCS features are also introduced.Using software-defined radio,WiFi signals are collected,and multi-domain features are extracted from modulation domain,time domain and frequency domain respectively.Finally,combining with random forest model,an identification decision is made for each frame transmitted by the device.In the real environment,the identification accuracy of 15 different kinds of WiFi devices can reach 98.08%,and the identification accuracy of 10 same-model network cards with smaller differences in hardware manufacturing can also reach 90.76%.The identification accuracy rate of all 25 devices reached93.19%.Finally,the prototype system of Io T wireless device identification is realized.By combining fingerprint identification technology based on link layer broadcast packet,link layer probe request and multi-domain physical layer features,the Io T wireless device identification prototype system is designed and developed,which realizes the discovery and identification of BLE and WiFi devices in the Io T scene.To sum up,this thesis uses common software-defined radio to collect wireless signals,extracts fingerprint features from data link layer and physical layer respectively,and realizes a prototype system of BLE and WiFi device identification.These research results can be applied to network access control,forgery attack detection and other fields.