Accurate Motion Tracing on Mobile Device and its application

Abstract

Motion sensors from mobile device, e.g., smart phone, smart watch, etc., can tell how the device moves. This project aims at achieving accurate motion tracing on mobile devices via motion sensors and investigates how this capability can be used to develop useful applications in practice. Accelerometer is the major motion sensor to be used in this project. Some applications in the market which use motion sensor only focus on large-scale and coarse-level measurements. Hence, we instead target at a fine-grained distance measurement using motion sensors. We envision that mostly people do not bring ruler on street so they could use it as a ruler to estimate length of physical objects. Moreover, we imagine when a person wears a smart watch, the motion sensors on the watch may be able to infer certain subtle gestures that user performs. Consequently an application that could reveal lockscreen password is created in this project. An accuracy of 97.4% of distance measurement is achieved. We also found that lockscreen password recovery is possible and has high successful rate. Within ten trials, ten of them successfully reveals password in top 10 results and two of them have the first rank. It proves a potential security problem of people wearing smart watch while entering password. It also proves that device-human interaction has many ways and possibilities to be explored. The project has smooth progress and is finished within timeline. Deep studies on motion sensing is proceeded and maintained in this project. Methodology of accurate motion tracing and application construction are explained in detail in this report.