Abstract: In view of the protection of personal information security of mobile phone, a user authentication scheme based on sensor information analysis is designed. The scheme uses the built-in accelerometer sensors and touch screen sensors to collect the users’data, analyzes the characteristics of telephone answering, sliding touch screen operation, standby state behaviors and carries on the classification and recognition separately by dynamic time warping algorithm. Because the single feature is limited to the specific condition and the stability of the identification result is poor, the multiple linear regression method is introduced to design the fusion decision model. The scheme does not require specific hardware device configuration, which can meet the requirements of users’ comfort in the actual scene and accurately identify the user’s control of the mobile phone.

Key words: mobile phone user authentication, acceleration sensor, touch screen sensor, weighted authentication, multiple linear regression model

摘要： 针对手机个人信息安全的保护，设计了一种基于传感器信息分析的用户认证方案。利用内置加速度传感器和触摸屏传感器采集用户数据，对电话接听、触屏滑动操作、待机状态等行为进行特征分析，运用动态时间规整算法分别进行分类识别。考虑单一特征局限于特定状态导致识别结果的稳定性较差，进一步引入多元线性回归方法构建融合判定模型，有效提高了个体区分度。该方案不需要特定的硬件设备配置，满足实际场景中用户使用舒适性要求，实验仿真结果验证了方案的有效性，能够准确辨别用户本人对手机的控制状况。

关键词: 手机用户认证, 加速度传感器, 触摸屏传感器, 加权认证, 多元线性回归