Research On Identity Authentication System Based On Smart Wearable Device

With the rapid development of mobile Internet technology,smart terminal devices have became indispensable tools in people's lives.In recent years,smart wearable devices such as smart bracelets,smart glasses,and smart running shoes have appeared in our vision,and have gradually became a new preference for the public.The characteristic of smart wearable devices that collecting users' biometric information conveniently could promote the research of biometrics.This thesis studies the identity authentication technology of mobile Internet,and takes the advantages of the smart wearable device easy to carry and conveniently collecting human biometrics,then proposes a SVM-based gait authentication scheme and a DTW-based PPG authentication scheme for the two scenarios of the user during walking and steady state.Next we implements identity authentication system prototype which based on the smart wearable equipment and combined the unified identity authentication framework FIDO UAF widely used in the industry.The SVM-based gait authentication scheme uses the acceleration sensor on the smart wearable device to collect gait acceleration data during normal walking of the human body.Then,this thesis performs smoothing denoising,normalization,scalar synthesis,etc.on the acceleration data and extracts time domain statistical features and frequency domain features in gait mode.And we further use the principal component analysis method to optimize the characteristics of the frequency domain.In this thesis,the feature combination with the highest accuracy in the time domain and frequency domain is obtained through experimental comparison and analysis,which is used as the gait feature template of the user.Next,support vector machine(SVM)is used as the classification algorithm to realize the authentication of human gait features.In the experiment,the gait acceleration data of 11 test users were collected,and the scheme was verified to obtain more than 96% of the authentication accuracy.The DTW-based PPG authentication scheme collects the PPG signal of the human body under steady state through the PPG sensor of the smart wearable device.The steady state here means that the user is in a relatively stable state,not in strenuous exercise or emotional excitement.Next,the signal is subjected to stability sequence detection,and extract the stable feature sequence representing the human skin tissue structure in the PPG signal,then we uses the DTW algorithm to perform offline training on stable feature sequences and filter out the optimal feature sequence template.In this thesis,the maximum DTW distance of the feature sequence is used as the threshold to determine whether the user has passed the certification by calculating the DTW similar distance between the test sample and the feature sequence template.In the experiment,the PPG signal data of 10 test users was collected and the scheme was verified to obtain more than 92% authentication accuracy.FIDO UAF is an online fast unified authentication framework proposed by the FIDO Alliance to address the serious fragmentation of devices caused by the wide variety of mobile terminal authentication.This thesis designs the authentication system based on smart wearable device and combined with the FIDO UAF framework.And this thesis has detailed design and description of the registration,certification,and cancellation processes in the system.In the end,we integrate the SVM-based gait authentication scheme and the DTWbased PPG authentication scheme described in the text and make it truly applicable to the actual application scenario.