MoViSign: A novel authentication mechanism using mobile virtual signatures

Signatures are a popular means of authentication. They, ideally, are unique to a person and, therefore, hard to copy. Two common types of signatures are traditional “pen-on-paper” signatures and electronic signatures. Electronic signatures are done on a custom-built electronic board using a stylus. The electronic board records information such as pressure, acceleration, and angle of the stylus in addition to the standard signature “image”. This information is used by a signature recognition system to differentiate authentic signatures from fraudulent ones. Both types of signatures have their drawbacks. While traditional “pen-on-paper” signatures are prone to forgery by expert humans, electronic signatures require a special electronic board to record the signature. In this project, we design and implement a novel mechanism of authentication, called mobile virtual signature, using mobile phones. Modern mobile phones such as the Apple iPhone and the Google G-phone are equipped with multiple sensors including accelerometers, orientation sensors, temperature sensors, proximity sensors, and multi-touch sensors. In our authentication mechanism, a person authenticates by signing “in the air” with the phone. The sensors on the phone, primarily the accelerometer and the orientation sensor, continuously record various parameters such as position, acceleration and angle. This information is then used by a Machine Learning program running on the mobile phone to recognize the signature. The authentication mechanism has several advantages. First, just like the traditional signatures, mobile virtual signatures stay unique to each person and can be changed if necessary. Second, it makes forgery hard as the signature is made “in the air” and without producing any tangible output