基于步态特征的移动平台持续认证方案

Mobile platform continuous authentication scheme based on gait characteristics

智能手机的普及使其对安全性有极高的要求,但传统的一次性认证方式,无法持续保证设备安全。针对此问题,提出了一种基于步态特征的持续认证方案,以实现对当前访问者的判别。所提方案采用移动平均滤波、基于阈值的有效值判断等操作降低噪声干扰,使用模板截取实现信息的最大化利用,并提出了最佳时域与频域特征组合以降低用户信息对存储空间的需求,通过支持向量机(SVM)实现身份认证功能。经实验测试,所提方案可达到对访问者身份的有效认证。

The popularity of smart phones renders people extremely high requirements for safety. But the traditional one-time authentication method can’t continuously guarantee the security of equipment. To solve the problem, a continuous authentication scheme based on gait characteristics was proposed to realize the identification of current visitors. Moving average filtering, threshold-based useful information interception method and other operations were adopted to reduce noise interference. Template interception was used to maximize the utilization of infor- mation, and an optimal combination of time domain features and frequency domain features were proposed to re- duce the storage space requirement of users’ information. Finally, the support vector machine realized the identity authentication function. Experiments show that the proposed scheme can effectively authenticate the identities of visitors.