Biometric authentication mechanisms are excellent alternatives to often inconvenient interaction authentication methods such as PIN numbers in mobile devices. This research introduces the new concept of using gait signature metrics for biometric authentication. This procedure verifies each subject using only acceleration. We first use a single wireless sensor device to collect data on subjects' gait patterns. By dividing each gait cycle into an Acceleration Phase and a Deceleration Phase, we derive seven periodic and characteristic gait signature metrics. Gait signature metrics can be classified as acceleration metrics, deceleration metrics, and ratio metric. Acceleration metrics represent a degree of dynamic activity when heel-strike actions and mid-stance actions occur, whereas deceleration metrics measure a degree of dynamic activity when mid-stance actions and successive heel-strike of other foot. The last metric, ratio metric, present the relationship between the acceleration metrics and the deceleration metrics. Using the gait signature metrics, we succeeded in differentiating each subject with 100% accuracy.