GPS vs Barometric Pressure Sensor altitude tracking
Summary
- GPS signals cannot always be obtained (especially indoors)
- Altitude can therefore not be measured, which results in inaccurate calorie expenditure estimations
- Barometric Pressure Sensors measure atmospheric air pressure to determine the current vertical position
- Accelerometer could compensate measurement errors within pressure sensor readings
- A combination of GPS, pressure sensors & accelerometer would be ideal
As you already know, GPS is not very accurate when it comes to altitude estimation. Especially in urban areas, forests or even just on a cloudy day, it is either hard to obtain a GPS signal at all or to achieve a reasonable accuracy (cf. Sigrist, Copping & Hermy, 2010; cf. Wing, Eklund & Kellogg, 2005)
Why is this important? To answer this, I’ll show you a measurement I conducted, comparing a barometric pressure sensor with a GPS signal obtained by a smartphone. For a better understanding of how the parking lot looks like, I found some similar pictures I can use for this blog. The second picture shows the parking lot from inside with the mentioned ramps that I walked down from the outside, the third from outside:
So, as you can see, this graph shows several events which need some explanation:
Why is the GPS altitude tracking data so different to the barometric pressure sensor data?
In the first part of the measurement, you can see that GPS is varying, but not as much as during the second and third part. This is, because the first few minutes were measured in an area with no trees, buildings or other obstacles nearby. When walking towards the parking lot, the accuracy in altitude measurement decreased. Inside the parking lot, it was not possible to obtain a GPS data point every second, due to the roof that blocked the signal. The four story parking lot has no walls, so sometimes it was possible to obtain a signal, but the accuracy was far off.
A simple (very simple) visualization for the cause of inaccuracy in GPS recordings in urban areas can be found to the left. The position is calculated from the time the GPS signal needs from the satellite to the device and back. Due to reflections from walls & windows, the time needed to return becomes longer and therefore results in a wrong conclusion of distance. (cf. Hahn & Powers, 2005) A barometric pressure sensor always measures the actual atmospheric air pressure and therefore is able to estimate the vertical position. With current technologies this is possible down to tolerances of smaller than +/-1 meter. A barometric pressure sensor cannot detect the horizontal position! The aim is not to replace GPS with pressure sensors, but to enhance the vertical position estimation by adding barometric pressure sensor data to the equation.
Gust of wind — how to take care of environmental influences affecting the consistency of the measurement
As you can see in the graph, there is a high peak for about four seconds. This was caused due to a strong gust of wind, which resulted in an air pressure change and concluded in a change in altitude of about four meters. This altitude change, of course, never happened and in cases like this, pressure sensor recordings alone would be inconclusive. The problem is, that this data is not only used for altitude tracking, but also to derive calories more accurately. So in addition to the error in altitude, it would also result in an error in calorie expenditure estimation. To avoid this, my suggestion would be to look at the data recorded done by the accelerometer within the same time frame. If no significant increase/decrease in acceleration can be observed, the measurement of the pressure sensor is inconsistent and can therefore be ignored. Oh, and by the way: I have no idea, why the GPS signal shows this strange altitude changes — the area where I walked at this particular time, was completely flat. So, even with the wind influence, the air pressure sensor would deliver more accurate results than the GPS module.
No pressure offset — altitude values at starting and end point are exactly the same
Now, this is impressive. I thought, the sensor would at least experience a little drift, which would result in an overall offset of about one meter. In other words: Start and end point of my measurement was exactly at the same location — and as you can see in the graph, the green line stops measuring at the same altitude level as it started. ‘Shouldn’t this be the case anyway?’, you ask — yes, it should be, but in general a drift can be expected. The solution would be a compensation of this error, which is hard to achieve but to a certain degree possible. What you see here in the graph is raw data which is not manipulated in any way!
GPS dead zone
— no signal, means no GPS data, means no altitude information
As I stated before, the parking lot had no walls, but a roof (obviously) that blocked the signal and therefore the phone was not able to obtain the current location. As soon as I got out of the parking lot, the GPS signal was received again. However, with the barometric pressure sensor, I was able to constantly measure the current altitude. If you take a look at the graph again, it is even possible to differentiate the floors.
Thanks for reading!
Follow me for new stories on WordPress or:
twitter.com @allaboutsensors
WordPress.com
https://youtube.com/@allaboutsensors5631
References
Journals
Hahn, J. H., Powers, E. D. (2005). Implementation of the GPS to Galileo Time Offset (GGTO). Proceedings of the 2005 IEEE International Frequency Control Symposium and Exposition, 2005, pp. 33–27. http://dx.doi.org/10.1109/FREQ.2005.1573899
Sigrist, P., Coppin, P. & Hermy, M. (2010). Impact of forest canopy on quality and accuracy of GPS measurements. International Journal of Remote Sensing, 20(18), pp. 3595–3610. http://dx.doi.org/10.1080/014311699211228
Wing, M. G., Eklund, A., Kellogg, L. D. (2005). Consumer-Grade Global Positioning System (GPS) Accuracy and Reliability. Journal of Forestry, 103(4), pp. 169–173.
