Ciholas is often asked, "how accurate is the RTLS evaluation system that comes with the DWUSB?" The answer, as with most things in engineering, is it depends.
Typical CUWB System
In a typical gym installation, where anchors can be placed with good geometry and no occlusions (line of sight visibility from tag to all anchors), the system is approximately ±10cm laterally and ±20cm vertically. This is for an installation with ten anchors and no significant occlusions. The CUWB system can produce this level of accuracy on any single tag beacon with about 97% reliability. As the tag moves the relative geometry to the anchors changes and will impact the overall accuracy at some locations.
The gym installation described above is shown in our demonstration videos:
Achieving Even Better Accuracy
As one might expect, better accuracy can be achieved through the use of location averaging. Averaging of locations has the same net effect as adding more anchors to the system.
The Archimedes release of the CUWB system is a demonstration system and doesn't represent the best accuracy that can be achieved for a given application. It is possible to conceive a system with a large number of anchors, tags with high beacon rates, and averaging of results over a long period of time. Such a system would potentially produce measurements with millimeter level accuracy. This concept is similar to techniques used for measuring plate tectonics where accuracy requirement are expressed in millimeters per year. In plate tectonics, GPS receivers are used with only a few meters of accuracy per measurement and averaged over large sample sizes and long periods of time.
Taken to the extreme, an absurd example, would be placing anchors at 10cm intervals on a gym ceiling. The system would be extremely precise, producing sub-millimeter results. However, it would require about 70,000 anchors to build!
To build a slightly more reasonable system, one might require an anchor every meter. The typical gym would then need approximately 700 anchors and would achieve an accuracy of less than 1cm.
Full assessment of the accuracy a real world application can achieve takes detailed examination of anchor placement, tag beacon rates, response time, etc. The fundamental trade off is accuracy versus response time and/or anchor count.
Accuracy vs Precision
It should also be noted that there is a difference between accuracy and precision. In reality, the numbers discussed above are precision (how repeatable the position data is) and not necessarily accuracy (how true the numbers are to the actual location). Consistent error will negatively impact accuracy but not precision. For example, errors in anchor survey, antenna group delay variation, or many other things can introduce accuracy errors. These same issues would result in a system that could still be very repeatable. A precise, but inaccurate, system can be made accurate by warping the measured data to fit the desired result. Note, the gym videos linked above are NOT warped or fit. The results achieved in these videos are direct results mapped onto accurate representations of the floor lines.