摘要
针对目前生物识别技术在穿戴式设备上应用的缺陷,提出一种可应用于可穿戴设备上的生物识别方法。利用300 k Hz^1.5 GHz的电磁波在人体通信信道传输中产生的幅度衰减特性曲线作为生物特征。为了验证此方法的可行性,首先,利用矢量网络分析仪测量生物特征;其次,提取数据的梯度,使用支持向量机进行分类器模型训练和测试。验证结果与直接对采集的生物特征进行分析的方法对比,引入梯度的分析方法使得正确识别率从90.45%提高到94.54%,等错误率从0.95%降低到0.14%,接收者操作特征曲线下面积从0.997 1增加到0.999 9。因此,基于人体通信的身份识别方法可为穿戴式设备的身份认证系统研究提供一种方法。
Due to the traditional biometric identification methods are not suitable for wearable devices, this paper proposed a novel type of biometric identification for wearable devices. The propagation characteristics curve of the HBC channel at the wrist within the frequency range of 300 kHz - 1.5 GHz were as the biometric trait. To investigate the feasibility of HBC identification, firstly,it measured the biometric trait by vector network analyzer. Moreover, it extracted gradients from measurement da- ta, and analyzed the gradients by support vector machines （SVM）. The performance of the gradient analysis method compared with the method of direct analysis the biometric trait, and result shows the correct identification rate （CIR） increased from 90.45% to 94.54% ,the equal error rate （EER） reduced from 0.95% to 0.14% ,the area under receiver operating characteristics curve （AUC） reached from 0. 997 1 to 0. 999 9. Therefore, biometric identification based human body communication can put forwards an idea for authentication system of wearable devices.
出处
《计算机应用研究》
CSCD
北大核心
2017年第4期1141-1144,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(U1505251
61403366)
广东省省级科技计划项目(2015A020214018)
深圳市技术开发项目(CXZZ20150505093829778)
关键词
穿戴式设备
人体通信
生物识别
梯度
支持向量机
wearable devices
human body communication（HBC）
biometric identification
gradient
support vector ma- chines
