【作者】 高山；

【导师】 刘明；

【作者基本信息】 电子科技大学 ， 计算机系统结构， 2017， 硕士

【摘要】 移动触摸设备诸如移动智能手机、智能可穿戴设备,ATM机触控屏等,已经融入进了人们的生活,改变了人们传统以来的生活方式。这些设备通常都存储有用户的隐私信息,例如地理位置信息,购物喜好信息,银行支付信息,身体健康信息等,不再单单只用于电话与通信。这些隐私信息的存在使得移动触摸设备变得越来越私密化,即设备与用户存在绑定关系,与此同时保护设备上隐私信息的安全也就变得愈加重要。我们根据设备在识别认证用户过程中所基于信息类型的不同,将应用在移动触摸设备上用户身份识别技术分为以下两个类别:根据用户所知信息的身份识别如:密码,口令;以及根据生物本身特征的身份识别如:指纹,虹膜等。但是这两种认证方式均存在不足:前者用户所知的信息容易被用户遗忘,且可能被恶意攻击者窃取;而基于生物特征的认证方式需要在移动触摸设备上集成造价昂贵的采集装置且要求采集信息精度高。基于以上背景,本文将重点着眼于触摸设备上的身份识别问题上:在触摸设备上如何实现成本低,精度高,方便简洁的身份识别。为此,我们发现人体生物电容为一种新型的可用于用户身份识别的生物特征,基于对此特征的研究,我们将该特征与人体行为特征相结合,发现两者结合后的特征具有唯一性和可采集性的特点,我们将两者结合而成的特征称为——人体差分电容,我们基于这两个特征的结合设计了一个新的用户身份识别系统。在系统设计过程中,我们采用了功耗极低的传感器芯片来采集待识别用户的人体差分电容。并通过我们设计的算法:基于滑动串口周期划分与基于快排周期回归合并算法,对原始数据进行预处理,此后我们提取出了四维向量用来描述人体差分电容,并基于该特征向量我们选用多项式核函数SVM分类算法进行用户身份识别验证。从我们设计的身份识别系统测试结果可知系统具有识别准确率高、低功耗、速度快等特点。其中识别准确度平均高达87%,基于事件触发的工作模式使得系统的总能耗稳定在μA级别,平均识别耗时仅为1.2s左右。该系统能够满足在移动设备上作为外接设备继承的需求,这样便可以在移动触摸设备上进行大规推广普及。更多还原

【Abstract】 Mobile touch devices such as smartphones, smart wearable equipments, ATM machine touch screen and so on, changed the traditional way of human beings’ life.These devices usually store the user’s privacy information,such as location information,shopping preferences information, bank payment information, physical health information and so on, no longer just for telephone and communication. The presence of these privacy messages makes mobile touch devices more and more private, the device has a binding relationship with the user. At the same time, to protect the privacy of equipment on the security of information becomes more important.We divided the mobile touch devices into following two categories by the type of information to authenticated user the devices based on: The user known information such as cryptogram, password; as well as identity based on the characteristics of the biology itself, such as fingerprints, iris and so on. However, these two kinds of authentication methods are inadequate: for the first method,the information known to the user is easily forgotten, and may be stolen by a malicious attacker; the second method need expensive equipment for feature acquisition.Based on the above background, we will focus on the issue of identification on touch device: how to achieve a low cost, high precision, easy and simple identification system. At last, we found that human bio-capacitance as a new type of biometrics that can be used for user identification. This features are unique and collectible, we combine human bio-capacitance and human behavior which called human-differential-capacitor,based on these new feature we designed a new user identity system. The original data is preprocessed by our own algorithm. Then we extracted the four-dimensional vector used to describe the human body differential capacitance and used SVM algorithm to do the user authentication.This system is a high-security, low-power-consumption (μA level), high accuracy(87%) and low time consumption (1.2s)recognition system, and meets the mobile devices add-on platforms requirements.更多还原