Sensor dropout and stale values
Make the policy for missing input explicit before stress testing.
Research note · November 14, 2026Research question
What should happen when a controller receives no new measurement?
Method
- Define a controlled sensor stream and interrupt it for known durations. Record whether the consumer holds the last value, substitutes a default, stops or reports an unavailable state.
- Preserve the age of the last valid measurement. A repeated numeric value should not become a fresh observation merely because a polling loop reads it again.
- Restore the stream and inspect recovery. Include a case where the physical state changed during the outage so an old value becomes actively misleading.
Records to retain
- Last valid measurement and time
- Dropout interval
- Fallback policy
- Recovery behavior
Source basis
Timestamp provenance distinguishes a newly observed measurement from a reused value. A simulation protocol should expose that distinction so the tested failure policy is clear to a reviewer.
Interpretation & limits
No fallback is universally safe. Its suitability depends on the system and hazard analysis. This note proposes test cases rather than choosing an operational policy for an actual device.
Next experiments
- Test short and long outages independently.
- Record the first valid post-outage measurement and the controller’s reaction.
References
Technical reference
Implementation details supporting this note, drawn directly from the maintained documentation. These are technical references, not additional papers or new experimental results. Expand a topic to inspect its method, fields and limits.