Sensitivity — what to measure first
Not "does it work at the assumed values" — at what value does it break? Each parameter is swept across its plausible range with everything else held still. Priority = swing × (1 − confidence): influence weighted by ignorance.
Ranking — effect on coverage loss
| # | parameter | swing | conf | priority | |
|---|---|---|---|---|---|
| 1 | ble.packet_delivery_ratio swept 0.3 → 1.0 fraction |
58.3 pts | 0.2 | 46.7 | |
| 2 | ble.path_loss_exponent swept 2.0 → 4.0 |
58.3 pts | 0.25 | 43.8 | |
| 3 | gateway.scan_duty_cycle swept 0.2 → 1.0 fraction |
63.3 pts | 0.4 | 38.0 | |
| 4 | ble.tx_power_1m_dbm swept -70.0 → -55.0 dBm |
41.7 pts | 0.6 | 16.7 | |
| 5 | ble.receiver_sensitivity_dbm swept -100.0 → -85.0 dBm |
35.0 pts | 0.7 | 10.5 |
A well-known parameter that matters is less urgent than a guess that matters. A parameter with zero swing comes off this list — retiring one for the right reason is as useful as promoting one.
Breaking points — coverage loss passes 25%
| parameter | breaks at |
|---|---|
| ble.packet_delivery_ratio | 0.3 |
| ble.path_loss_exponent | 3.0 |
| gateway.scan_duty_cycle | 0.2 |
| ble.tx_power_1m_dbm | -70.0 |
| ble.receiver_sensitivity_dbm | -90.0 |
A breaking point is a statement about the parameter, not the metric — "how bad can this get before we care" becomes a value a pilot can go and check against a real tag on a real site.
The report, as it would be exported
==========================================================
Sensitivity of coverage_loss_pct — scenario 'Yard A'
[ ASSUMED — NOT MEASURED ]
==========================================================
baseline coverage_loss_pct 13.33
ble.packet_delivery_ratio (swept 0.3..1) 58.33
ble.path_loss_exponent (swept 2..4) 58.33
gateway.scan_duty_cycle (swept 0.2..1) 63.33
ble.tx_power_1m_dbm (swept -70..-55) 41.67
ble.receiver_sensitivity_dbm (swept -100..-85) 35
* Under the assumed profile, ble.packet_delivery_ratio is the parameter worth measuring first: it moves coverage_loss_pct by 58.3 across its plausible range and its current value is little more than judgement (confidence 0.2)
* Under the assumed profile, coverage_loss_pct passes 25 once ble.packet_delivery_ratio reaches 0.3
* Under the assumed profile, coverage_loss_pct passes 25 once ble.path_loss_exponent reaches 3
* Under the assumed profile, coverage_loss_pct passes 25 once gateway.scan_duty_cycle reaches 0.2
* Under the assumed profile, coverage_loss_pct passes 25 once ble.tx_power_1m_dbm reaches -70
* Under the assumed profile, coverage_loss_pct passes 25 once ble.receiver_sensitivity_dbm reaches -90
----------------------------------------------------------
Evidence: ASSUMED — NOT MEASURED
No parameter behind this has been measured. The figure describes the assumptions, not the hardware.
Measure these for this report to become evidence:
- ble.packet_delivery_ratio
- ble.path_loss_exponent
- ble.receiver_sensitivity_dbm
- ble.tx_power_1m_dbm
- gateway.scan_duty_cycle
Profile: assumed-1.0.0
Generated: 2026-11-12T23:50:22+00:00
==========================================================