Using topography to aid cell phone geolocation

In daily life, people demand accuracy of the Global Positioning System (GPS) receiver. The current problem of GPS on mobile phones is that it is not available in areas such as urban, natural canyons, forests, and indoor environments. Several methods have been developed to obtain more accurate position estimation over the past years. The received signal strength (RSS) and time difference of arrival (TDOA) are the main approaches to use available mobile signals and errors around 4 ~ 12 dB and 10 ~ 60 meters, respectively. Another approach to make a better performance of the sensor is to use radio frequency identification (RFID) with indoor Wi-Fi. A new method from our group shows that using magnetic field intensity maps based on interval analysis can perform better than the RSS, TDOA and RFID and reduce error for geolocation in some areas where GPS is not accessible. In our study, we develop a novel algorithm where sensor measurements on the cell phone are used to construct the topographic maps and aid cell phone geolocation which focuses on the angles of inclination in user\u27s pathway when GPS is spotty. This can be particularly useful on uneven terrain outdoors. For sensor characterization, we use application in android operating system of smartphone by name of sensor stream IMU+GPS. The sensor stream allows for users to observe, select or record the current values of various measurements such as accelerations, angular rates (gyroscope), magnetic fields, GPS position and received signal strength indication (RSSI) in 3-dimensional coordinate system. We firstly develop algorithms of fast fourier transform and low pass filter to find the accurate vertical acceleration measurements which impact values are corresponded to step occurrences. Before analyzing position estimation, we use the relationship between the stride length and stride interval and the methodology of detecting peak values to find the user\u27s step. In order to reduce uncertainty and find the user\u27s walking direction in our navigation system, we apply the Kalman filter and rotation matrix. We then develop optimization algorithms to bound the local position estimation into small 2-dimensional intervals using the interval analysis and dynamic estimation. After transforming the history of gravitational vectors to a fixed local-coordinate frames, we are able to construct a topographical map of pathway. We test our methodology in controlled conditions on an instrumented treadmill and also outdoors where GPS is available. We then use our topographic mapping to augment the results from pedometry and magnetic mapping to obtain better geolocation