we are working with 8 DWM1000 integrated in an anchor device and several tags integrated in small quadcopters.
The 8 anchors are approximately on the vertices of a cube shape, the height is 7,5 m for the upper ones and the bottom ones stay at 45 cm above the ground. We have LoS and an empty space without metal parts inside and around it.
We have experimented and fine tuned different implementation of TDoA algorithms and now the calculated/filtered position in static test has around 3-5 cm standard deviation error.
The st.dev. of the single TDoAs is between 4 and 9 cm.
Now we want to improve the hw side of the system and we think we really need to focus on the antenna part.
Do you think its the right part to put more effort for improving the precision?
Reading some recent post here and on other site I’ve found out several reports about different antennas used with the DWM1000 with improved overall performance.
What antenna can I try to improve the system precision ?
How does your accuracy vary with location in the cube?
Changing antenna isn’t generally going to reduce position noise but it can reduce bias.
If you are getting the same accuracy at all locations then changing the antenna isn’t going to make any improvement. If you are getting different accuracies at different locations or a position dependent bias in location then changing antenna may help.
e.g. when you are near the sides of the cube do the positions all tend to be slightly off in a direction towards the middle of the cube?
How about orientation dependency on the quad copter? If you rotate the quad copter about the tags antenna do you still get the same position?
The solution is to move the antenna away from the circuit and make things as symmetrical as possible. On the plus side it sounds like you only need a 90 degree field of view from the antennas in the anchors, that should make things easier.
The quadcopter could be trickier since by definition the direct path to at least some of the anchors will be blocked by the body of the quadcopter.