Relation between T-reply length and Measurement errors in Single-Sided Two-Way Ranging (SS-TWR)

Hello everyone, I was conducting some experiments using the ex_06b_ss_twr_resp and ex_06a_ss_twr_init example codes.

I found something really unexpected. As far as I know, in SS-TWR, as the time reply increases, the effect of the clock offset error on the distance estimation also rises. Therefore, I would expect that when I changed the time reply from 300us to 4000us, the mean error would rocket. However, what the outcomes showed was very different.
Below, there is a figure comparing my measurements. The mean error was calculated considering 15 samples.

The configuration I used was: channel 2, PRF 64 MHz, 6.8Mbps, Preamble 64, SFD 8, the default antenna delay for TREK1000 (I am using EVB1000s from a TREK1000 kit), the function dwt_getrangebias from the TREK1000 source code and the two functions that increase performance in a preamble length of 64.

I suspect that since the initiator end of those example codes uses the carrier integrator to correct the range estimate calculation for the clock offset of the remote responder node, the increasing clock offset error becomes kind of negligible. Am I right?

Another question, are those results displayed in the figure good enough for SS-TWR systems?

Thank you very much for your time!

For 4000us Trep you have a less acurate results than 300us, while after 40m the reverse.
Accuracy should also be consistant, and that is not happening either.
Could you say something about the environment (Indoor / outdoor / obsticles) and teh height the nodes are positioned? What you experience could well be because of reflections , see section 3.4.2 Multipath due to ground-bounce in attached Application note.
I don’t think increasing the Trep will improve your accuracy fluctuation.

Leo APS006_channel_effects_on_range_accuracy_ver1.03.pdf.pdf (1.1 MB)

1 Like

Thank you for your reply.

The nodes are positioned at a height of 1.72m. There is line-of-sight and the environment is a corridor of about 3.5m width and 3m height. Now that you said, that inconsistency in the results must have been due to the reflections caused by the environment. I am going to read the PDF you sent me.

I appreciate your help.