smart user authentication through actuation of daily activities leveraging wifi-enabled IoT

发表在Mobihoc‘17,是由陈莹莹团队研究的成果。


目的:通过用户的日常动作(walking,stationary activities)认证用户。

主要贡献:
1)通过无线信号与动作之间的关系,获取一个人独特的人体特点和行为特点,进而通过日常动作认证这个用户。
2)利用CSI的振幅和相位来描述用户的独一无二的动作。
3)利用深度学习方法来侦测用户的日常动作,并且防止非法用户攻击(这个点比较好)
4)在两类实验环境(lab and apartment)分别获取94% and 91%的认证精度。

代价:
1)一对设备

2)5个月的时间

smart user authentication through actuation of daily activities leveraging wifi-enabled IoT

STE(short time energy): 短时能量

smart user authentication through actuation of daily activities leveraging wifi-enabled IoT

效果展示:

smart user authentication through actuation of daily activities leveraging wifi-enabled IoT


针对一个波峰,如果多个波峰,这个方法会存在一定的问题。我还在思考是否可以在继续改进,提高侦测动作起始点的精度和稳定度。


subcarrier selection:

****to eliminate the negative effects caused by the unstable subcarriers, a covariance based scoring function is defined to assess each subcarrier's correlation level with its neighboring subcarriers as follows:

smart user authentication through actuation of daily activities leveraging wifi-enabled IoT


smart user authentication through actuation of daily activities leveraging wifi-enabled IoTsmart user authentication through actuation of daily activities leveraging wifi-enabled IoTK:表示相邻子载波的数量

i,j:表示不同的子载波

效果如下:

smart user authentication through actuation of daily activities leveraging wifi-enabled IoT

我的思考:
我一直倾向于保留敏感的子载波,移除稳定的子载波。本文的工作则是移除不稳定的子载波,保留稳定的子载波。这一点挺有意思的,引起我的思考。哪一种说法接近真实的理论呢。还需要探讨和分析。





我个人观点:
该工作还是没有跳出我们常用解决动作识别的技术和方法,只是重新组合换了个说法,所以创新性还是不太突出。
“ we show that the proposed system is resilient to spoofing attacks when integrating the feature abstractions from the deep learning model with the SVM classifier”




smart user authentication through actuation of daily activities leveraging wifi-enabled IoTsmart user authentication through actuation of daily activities leveraging wifi-enabled IoT