UWB acceptance example — INVENTED teaching data
Guide: https://yenra.com/location-devices/

No hardware was measured. Ten static observations are scheduled at separate illustrative reference points in one local 2D frame. Each numbered window expects exactly one fresh position; duplicate reports are excluded by construction. Reference coordinates are assumed exact for this arithmetic example only. A real test must record truth uncertainty and time alignment.

CSV fields: window = observation number; dataset = invented; truth_x_m/truth_y_m = reference position in metres; estimated_x_m/estimated_y_m = reported position in the same frame; latency_ms = elapsed time from measurement to application receipt in milliseconds on an assumed common clock; status = returned or missing. Empty estimates and latency mean missing, never zero.

For each returned fix calculate hypot(estimated_x_m-truth_x_m, estimated_y_m-truth_y_m). Joint acceptance requires error <= 0.50 m AND latency <= 200 ms. Treat a missing fix as a failed scheduled observation for joint acceptance.

Expected result: 9 returned / 10 scheduled = 90% availability; 8 jointly accepted / 10 scheduled = 80%; 8 / 9 returned = 88.8889% conditional acceptance. The ninth fix has 0.80 m error and 260 ms latency. The tenth is missing. The tiny sample is an arithmetic demonstration, not a statistical performance claim.