关键词: 无线安全, 无线设备身份异常检测, 信道指纹, 信道状态信息, 局部离群因子

Abstract: With the rapid development of wireless technology and the broad coverage of wireless communication infrastructure, wireless network has become a necessity in people's daily life．However, the endless impersonation attacks have greatly challenged the security of wireless network．Encryption system is a traditional solution to resist these attacks, but its improvement of security is based on the increase of system production cost and computational complexity．At the same time, the continuous upgrade of cracking techniques also poses a threat to its security． This paper proposes a non-cryptographic wireless device identity anomaly detection system．The system uses fine-grained channel state information (CSI) to generate channel fingerprinting．Such fingerprinting represents the physical properties of device and is therefore difficult to forge．The system uses fingerprinting to construct local device calibration files, and uses the local outlier factor (LOF) to verify devices’ identity．Considering the location sensitivity of channel fingerprinting, we provide a secure calibration file update method, which can avoid misjudgment of legal devices whose location has changed, and can also detect impostors who try to interfere with the update．Experiments showed that our system has a high accuracy rate of anomaly detection, which has a good application prospect in wireless network security．

Key words: wireless security, wireless device identity anomaly detection, channel fingerprinting, channel state information, local outlier factor