|Most smart phones include several MEMS sensors. The data they provide |
helps make location possible when GPS signals are compromised.
For example, accelerometers measure acceleration based on user movement. Generally, a three-axis accelerometer measures the change in user velocity position in a variety of cases, such as walking, running, falling, and vibration.
Gyroscopes measure rotational velocity. They also are implemented across three axes. Gyroscope measurements provide orientation. A combination of accelerometer and gyroscope measurements in three dimensions can be used to continuously track user position.
Magnetometers measure the earth’s magnetic field as well as any ambient magnetic influences. They are used for mapping and compass functions (potentially in conjunction with accelerometers). And, barometers measure the altitude based on air pressure changes and can be used to find a user’s position while moving up in altitude (Fig. 4).
Physical readings from the sensors are typically sent to the inertial measurement unit (IMU), which collects data from the inertial sensors, formats them, and delivers them to the sensor location subsystem—the inertial navigation solution (INS). Yet there’s a key difference in the positioning technique used by the INS compared to other location methods.
A-GNSS and Wi-Fi positioning and cellular positioning methods feature an absolute position computation process. The INS uses dead reckoning. It’s given a starting position (perhaps from an A-GNSS fix) and then uses sensor data such as speeds and angular direction to advance the position over time. Eventually, the uncertainties in the reported sensor data add up, and a new, accurate reference position is required to start the dead reckoning process over.
While GPS, Wi-Fi, and cellular technologies provide the user’s position on the map, sensors can provide an idea of how the mobile device itself is moving—whether it’s being turned, set on a table, or thrown. This fine-grade information can be very useful for enabling applications such as gaming. It also presents a challenge.
If purely position calculation is of interest, then the positioning engine must be able to intelligently filter out extraneous effects such as the rhythmic motions of walking or running and holding the handset in different positions while talking. Different people have different usage profiles, so the sensor’s positioning engine must be able to robustly handle a variety of different scenarios.