Gait-based User Recognition For Wearable Devices

With the rapid development of the Internet,wearable devices gradually grow into the tens of thousands of households,such as wearable wristbands,smart watches and so on.The wearable device has powerful data collection and transmission capabilities,collecting user's heart rate information,acceleration information,gait information and location information;collecting user's daily communication text data,audio data and video data.The security of wearable devices is critical to users,and the common way of inputting the password can easily Iead to password leaks.When the user wears the wearable device,how to determine the device owner is very important for protecting the security and privacy of the device owner.This paper studies the user identification problem of wearable device,and we take the user's gait information as the element of user recognition.Although there are many existing gait-based recognition methods,almost all of them do not consider the energy consumption,or using expensive hardware to complete the recognition task and ignore the cost of the wearable device.On one hand,if we do not consider energy consumption and the wearable device needs frequent charging,consumers will become reluctant to use such wearable devices;On the other hand,using additional hardware equipment will increase the cost of wearable device.Human gait information has been used as a unique user characteristics for a long time.This paper is to achieve user recognition based on user's gait information.Our work in this paper consists of two parts:1.For multi-user device,the paper can use a multi-classification method,and we use Naive Bayesian classification method to achieve higher recognition accuracy and lower energy consumption.2.For single-user device,the paper proposed and implemented an online user recog?nition system that can build a complete recognition system without additional in?terference from the device owner.