| Access control system is an effective measure to ensure the safety of important places,and has become an indispensable part of security prevention.With the development of Radio Frequency IDentification(RFID),RFID-enabled access control gate has just been unfolding.However,the traditional RFID access control adopts the channel with barriers,which requires users to line up and swipe their cards,resulting in the congestion and low efficiency.In view of the above problems,open access control without barriers comes into being.Compared with traditional access control,open access control does not require the card swiping.Users can go through the channel without any operation and waiting,which greatly improved the efficiency.Unfortunately,existing open access control gates have lost the function of direction monitoring while bringing convenience.In additon,for the illegal intrusion detection under the continuous traffic of multiple users,existing solutions mostly rely on the high-cost dedicated sensors,and there is no lightweight algorithms can detect the intrusion of illegal people who follow the legal users.Considering above limitations,this paper utilizes UHF RFID technology to implement high-precision direction monitoring and lightweight illegal intrusion detection by extracting features from the backscattered RF signals,without any additional equipments.The contributions of this paper are summarized as follows:(1)For the problem of direction monitoring in open RFID access control,a slotbased dual-antenna model is proposed.By deploying a pair of antennas on one side of the channel,the model creates asymmetry features in the symmetric occasion.In the communication with the same tag,based on the inherent differences in timestamp sequence of different antennas,a slotted direction detection algorithm is proposed to realize the high-precision direction monitoring.(2)For the problem of illegal intrusion detection in open RFID access control,a model based on tag array is proposed.By deploying a tag array on the other side of the channel,the model extracts features from the whole array including RSSI entropy,Doppler frequency entropy,etc.Then,Random Forest is used to estimate the number of human objects in the channel in real time.Finally,compare estimated number of people with the number of legal tags which are read in the same period to realize the lightweight illegal intrusion detection.(3)The above two models are combined to build a prototype system to verify the performance of the models.Multi-dimensional experiments fully demonstrate the high reliability of the proposed models.Under the comprehensive evaluation,the accuracy of the slot-based dual-antenna sensing model is 99.83%,and the accuracy of the arraybased sensing model is 96.67%. |
