Understanding the effect of positioning on mobile sensor data and its impact on applications

Abstract

Most sensor based mobile phone applications like navigation systems, pothole detection, pedometers, traffic detection, etc. rely on the strict and stable positioning of the mobile phone at a particular place throughout their usage. Most of these applications also use accelerometer and orientation sensor data which is highly sensitive and susceptible to the external magnetic interferences, resulting in inaccurate information about the directions. Our aim is to study the usage of modern sensors like gyroscope and rotation vector sensor to improvise the ability of smartphones to track movement. And also, to reduce the effort of stable positioning of the phones by virtually reorienting it along the three axes. This will also address the interference problems associated with the orientation sensor. This is necessary because the data that we get from the sensors is with respect to the orientation of the phone, i.e., along its own x, y and z axes. But we would first rotate the phone virtually from its own axes to the actual axes by applying rotational mathematics. This would help many applications which use sensor data to become independent of the phone's position. We evaluate our results by comparing the original values and the calculated values of the sensor by calculating the correlation coefficient. We get the best results for the rotation vector sensor and conclude that it is the most reliable sensor which can be used to virtually reorient a smartphone.