Performance of gait authentication using an acceleration sensor

摘要

ICT and MEMS technology are becoming mature enough to raise interest in wearable sensors for creating various sensor-based services. This paper introduces a gait authentication method, a kind of biometric authentication, based on wearable sensors. Gait movement is both unique and unavoidable in everyday life. The pattern of acceleration values created by walking differs from one person to another. Therefore, we expect various new services can be created based on the gait data obtained by wearable sensors on a daily basis. We capture gait data from a wearable sensor and analyze personal gait characteristics on the basis of the data obtained. Each subject placed a wearable sensor on the ankle of the right foot. We propose and evaluate an authentication algorithm. The authentication algorithm must converge on the attributes that allow for reliable discrimination. Therefore, our authentication algorithm uses the traveling acceleration at the swing phase. Results confirm that setting the threshold value to 1700 yielded EER of 20%