Implicit authentication mechanism plays an important and unique role in addressing the collision between security and usability on intelligent mobile device.However

most of current studies are usually built on a single feature or action

while suitable for a particular action and scenes.To solve this problem

we proposed a multi-feature fusion based on implicit authentication scheme

which uses the unique feature

such as the location

environment

posture

gait

biometric and behavioral

during use device.In the scheme

the data

such as sensors

biological and behavioral data

are collected

multi-features are trained and extracted by using the support vector machine.Then

the multi-feature fusion model is designed

and the framework of implicit authentication is constructed for calculating the user confidence level.At last

personalized security policy is designed

and the scheme authenticates user continuously and transparently.Experimental results validate the effectiveness of the proposed scheme and balance the security and usability and energy consumption.