Minimum distance between anchors

[santiagocano298366]

Hello!
I would like to know if there is any minimum measurement in which the anchors should be separated
I am working with 4 anchors

[jremington]

The fractional error in position decreases as anchor separations increase, so very short anchor-to-anchor distances, like 1 m, would not make much sense.

[santiagocano298366]

My anchors are separated in a 3mx2.1m box and I am having a problem in which the values ​​are stable, but the values ​​start to rise
I am using code with 1 tag and 4 anchors in 2D
What could be happening?

[jremington]

Sorry, I can't much sense of the phrase "the values are stable, but the values start to rise".

[santiagocano298366]

The tag values are stable for a while, however, out of nowhere the tag values increase without moving position.

[jremington]

Consider adding some useful details, or examples.
Which values change, and by how much?
What is the behavior over a long period?

[santiagocano298366]

We have located the tag at the coordinates x:90 and:2.75. At the time of reading through the tag, it gives us values ​​close to that measurement, but after a while it begins to increase the measurements, reaching x:60-1.20 and:2.55-2.90. these values ​​for a while and then stabilize again to the extent that the tag really is,
The time in which this happens is not constant and happens arbitrarily

[jremington]

Are you implying that the tag is about 90 meters from the origin of the 3m x 2.1m box of anchors? If so, I would expect the results to be highly inaccurate, and unstable due to errors in distance measurements.

For highest accuracy, the tag should not be far outside the perimeter defined by the anchors.

[jremington]

I ran a simulation to obtain an estimate of the distribution of errors versus distance of tag to anchor origin.

The anchor box is 10m x 10m x 3m. 10000 trials were conducted for a tag placed within a box 40m on a side, corners +/- 20 from the origin of the anchors. Added to each distance measurement were random measurement errors with Gaussian distribution, standard deviation 0.1 m. Here is the result for 10-fold averaging of each location determination (y axis = position error, x axis = distance from origin):