I have been testing the MDEK1001 kit to determine the accuracy and performance of the kit.
Throughout my tests I have realised that the Z-coordinate for the tags’ position is very inaccurate/wrong.
I have arranged 4 anchors in a rectangle on the floor of a completely open space (so there should be no line of sight issues). I have manually configured the position of each anchor respectively, and setting the Z-coordinate to 0.0m. When I move the tag within the rectangle along the floor (so it is at the same height as the anchors) I get a Z-coordinate measurement from the tag of approximately 0.5m. I understand there is a +/-10cm accuracy of the measurements, so I would expect the Z-coordinate measurement to be 0.0 +/- 0.1m, not 0.5m?
The 1-D range estimate is ± 10 cm, to get a location (and from it deduce the z-measurement) the 4 ranges/with any error are combined in the Location Engine. Having anchors on same height will not give very good z, this is due to GEDOP effects. E.g. in x and y planes the anchor separation is ~number of meters … this then gives good intersection (of spheres) and smaller error, thus to get equivalent error in z the anchor separation in z plane should also be couple of meters at least.
Idea to try:
Place two anchors on the floor, and two on the ceiling (diagonals of the room should have same height). Let me know if the z improves
I have re-run the tests with different set ups for the anchors. For the first set up I tried your suggestion of having two anchors on the floor and two anchors high up near the ceiling. For the second set up all of the anchors were on the high up near the ceiling.
This massively improved the Z, making it accurate to approximately ±5cm, so thank you for this help!
However, although my Z improved, this seemed to actually decrease the accuracy of the X and Y (down to approximately ±15cm from ±5cm from the original set up).
Is this expected? Any ideas to optimize accuracy of all dimensions would be great!
