Door Lock Security Enhancement Based On Non-contact Sensing Of Hand Rotation

Fingerprint recognition,iris recognition and face recognition technology have been applied to intelligent access control system.However,these physiological information has been proved to be replicated and thus the security of access control system is destroyed.A large number of wireless network devices are widely deployed in the indoor environment.The network sensing environment makes it possible to design a new generation of intelligent access control system.This paper focuses on the research of non-contact intelligent door lock security enhancement technology based on traditional key unlocking action.The main work can be summarized as follows:Firstly,a hand rotation model based on channel state information is proposed.By modeling the physical characteristics of human movement and the changes of wireless signal physical characteristics,it is proved that the amplitude and phase of CSI contain the unique physiological characteristics of users after removing the static variables in the signal,which provides theoretical support for the safety enhancement of non-contact intelligent door lock based on the traditional key unlocking action。Secondly,a new way of user identity marking called Full Coding is proposed.In the deep neural network,Full Coding uses a fixed length sequence of digits to uniquely identify a user,and all digits in the interval [0,9] are used as label vectors.Using this method,when the number of users increases,the memory occupied by the vector marking user identity will not increase.When new users are added,the original deep neural network structure does not need to be reconstructed,which makes up for the shortage of updating cost of traditional coding methods.Thirdly,a deep neural network classifier based on attention mechanism is proposed.First,the user’s motion information is encoded into the hidden state of the encoder,and then the score vector is calculated through the attention mechanism,and the classification results are output combined with the hidden state of the decoder.This method can process the variable length input data,and the output result is the Full Coding result,so as to determine the user identity.Finally,the commercial Wi Fi device is used to deploy the door lock security enhancement system named Airlock based on user unlocking action,and the experiment is designed to evaluate the system.The results show that airlock achieves an average authentication accuracy of 96.5% in the original scene and 89.5% in the scene migration experiment.