Study On Physical-layer Authentication With Multiple Landmarks In Wireless Networks

With the development of mobile Internet and Internet of Things(loT),wireless network is threatened by attackers because of its open and mobile characteristics.The spoofing attacker with faked Media Access Control(MAC)address sends spoofing packets to obtain illegal advantages and further perform denial-of-service attacks.Physical(PHY)-layer authentication techniques exploit physical layer properties of wireless communications,such as the channel responses that varies with position to detect spoofing attacks.Although PHY-layer authentication scheme has less computation costs,the detection accuracy has to be improved.In this paper,a distributed authentication network based on Multiple-input Multiple-output(MIMO)technique and multiple landmarks is proposed to improve the authentication accuracy with sufficient user channel information.The authentication model based on the logistic regression does not relay on specific channel model,and has a general validity.On this basis,a centralized authentication scheme based on Frank-Wolfe algorithm is proposed.A distributed authentication scheme based on distributed Frank-Wolfe(DFW)algorithm is further proposed for the distributed authentication network to reduce the communication cost among landmarks.The experiment results show that the proposed authentication scheme can reduce the communication cost with high detection accuracy.For example,compared with the scheme based on Frank-Wolfe,the proposed scheme with 8 landmarks and 8 antennas reduces the communication cost by 28.9%with 2%false alarm rate and 0.1%miss detection rate.With user's identity document based on channel information,a distributed authentication scheme based on Distributed K-Means(DK-Means)algorithm is proposed to reduce the dependency on the labeled training data in DFW based authentication scheme,and further reduce the communication cost.The experiment results show that the proposed scheme reduces the communication cost with high spoofing detection accuracy.For example,compared with the DFW based authentication scheme,the proposed scheme with 8 landmarks decreases the communication cost by 51.6%with 99.9%authentication accuracy.