othmane00m / gpslogger

Automatically exported from code.google.com/p/gpslogger
0 stars 0 forks source link

Improvement: add SNR, EDOP and NDOP #8

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago
In assessing the precision of a GPS track for mapping, HDOP is useful but is 
only part of the story. (see http://www.outdoormaps.org/?p=138 for discussion 
and links).
------
First: HDOP (horizontal dilution of precision) is the quadratic mean of "EDOP" 
(easting DOP) and "NDOP" (northing DOP).
If we had both EDOP and NDOP, we could represent the DOP in JOSM as an ellipse 
instead of a circle (as in http://www.outdoormaps.org/?p=128).
Use case: You have two GPS tracks. On both, you went through Queen's road and 
King's road.
On Queen's Road, the two GPX tracks are 50 meters apart, one (track A) on the 
east, the other on the west (track B). Track B has both high EDOP and NDOP on 
all points. Track A has high NDOP but very low EDOP. So, for the purpose of 
mapping a road oriented north-south, track A is very good. You discard track B 
and map Queen's Road on top of track A.

On King's Road, the two GPX tracks are 50 meters apart, one (track A) on the 
east, the other on the west (track B). Both tracks A and B have high EDOP and 
NDOP on all points of King's road. You draw  King's Road in the middle, at an 
equal distance of tracks A and B. (In OpenStreetMap, you may add those tags: 
fixme=resurvey; note=averaged from 2 low-precision GPS tracks).

Editors like JOSM would show the DOP like an ellipse, not like a circle

I provide some working prototype code (in R) and alink to the maths on how to 
do this at http://www.outdoormaps.org/?p=138; nothing hard here, tell me if I 
can help.

(A further improvement would be actually to store the direction of the highest 
DOP: this may not always be north-south or east-west).

------
Second: DOP conveys only the information on the position of the satellite. As 
discussed in the above link, SNR is equally important. One or more satellite(s) 
with a very low SNR may signal reflections or signal that is so low that it 
becomes barely usable.
Mean SNR (accross all used satellites) is probably an excellent start.
In particular according to one scientific reference[1], it is useful in 
combination to the number of satellites. 
[1] Pigott, "Inferring Transportation Mode using Smartphone Sensor Data", 2011, 
p. 42
http://www.piggott.me.uk/blog/2011/08/25/inferring-transportation-mode-using-sma
rtphone-sensor-data/

Ref [1] present a use case. ANother one would be to have circles/ellipse in 
JOSM that would show the accuracy based on combining xDOP and SNR.

Original issue reported on code.google.com by mayeul.k...@gmail.com on 8 Jan 2013 at 6:03