ABSTRACT
Now a days, Smartphones have become the best companion of the human. We usually save all our important documents, files, even most private data, passwords of various accounts and applications in our Smartphone. Also, there are many applications that support various business activities, online shopping and bank activities, which all involves money. Generally, the security of a Smartphone lies with entry point authentication such as passwords, PINS or Graphics patterns. But this authentication schemes are performed only once while unlocking the device and is not performed again till the device is again locked. So these methods can be breached by simple attacks like smudge attacks, shoulder surfing etc. The main problem in the entry point authentication lies in the genuineness of the user after the successful authentication. Most of the users prefer simple passwords or pins so that they can be easily remembered and the device can be unlocked easily. So, this scheme is not fully safe. In this work we present an implicit continuous authentication scheme, which periodically authenticate the user in the background by passively analyzing their touch biometrics. Here we have used 2 touch biometrics swipe data and 3 classifiers-Nearest Neighborhood, Support Vector Machine and Deep Learning Neural Network in recognizing the legitimate user. The result shows the accuracy of each of the classifier for two different datasets.
