Specific Absorption Rate Based Identification System On Mobile Devices

Mobile devices, such as mobile phones, smart wearable dvices, have changed the traditional way of life. They are no longer used as telephone and network communication tools, but also used to sense users and collecting private informations, like GPS locations, preferences, healthy issues and so on. Modern usabilities make mobile devices become more and more prevalent and personal, which gain more attentions on the information protection on them.Based on the kind of the feature, human identification schemes on mobile devices can be classified into two ways:(1) “knowledge-based” user identification(password, token, etc.).(2) “biometric-based” user identification(fingerprint, iris, etc.). While being widely employed on off-the-shelf mobile devices, these schemes still have some shortcomings: known knowledge shared between users and devices has often been forgotten or stolen, while specialized biometric capturing platforms are always expenvise and user must meet high-level caputuring standards.So in this paper, we focus on the fundamental problem in human identification on mobile devices: how to accomplish a low-cost, high-accuracy and easy-to-use identification on mobile devices. For this purpose, we find an unique and easy-to-collect biometric Specific Absorption Rate(SAR), based on which, we have propose a human identification system on mobile devices. This system can sense users’ SAR in ultra-low-power-consumption mode, applies sliding-window-based period partition algorithm and quick-sort-based regression and merging algorithm on original sensing data to etract four-dimension feature vecotor, and use polynomial kernel based SVM to classify and recognize users’ feature vectors.Result shows that this system is a high-security, low-power-comsuption(at ?A level), high-accuracy(87% with only 17 learning samples) and low-time-consumption(1.2s) recognition system, and meets the mobile devices add-on platforms requirments. So that it can be embedded and widely used on the mobile device to recognite users at a high accuracy.