S2. The sensor is returning readings but they are not correct.
S3. The sensor is not returning readings.
To the pilot (or automation) using the sensor there are only two states:
U1 = S1 + S2.
U2 = S3.
With two you can be in 7 states.
S1. Both sensors are returning correct readings.
S2. Both sensors are returning the same incorrect readings.
S3. One sensor is returning correct readings and one sensor is returning incorrect readings.
S4. Both sensors are returning incorrect readings which do not match.
S5. One sensor is returning correct readings and one is not returning readings.
S6. One sensor is returning incorrect readings and one is not returning readings.
S7. Neither sensor is returning readings.
The states the pilot (or automation) sees are
U1 = S1 + S2
U2 = S3 + S4
U3 = S5 + S6
U4 = S7
This is an improvement. If you get readings that agree (U1) or just one reading (U3) you can be more confident that the reading is correct than in the case of only one sensor.
Here are some numbers.
If the sensor works fine 98% of the time, but 1% of fails in a way that gives false readings, and 1% of the time fails in a way that gives no readings, then when the pilot gets a reading on the single sensor plane there is a 98.99% chance it is right.
On the two sensor plane when the pilot gets a reading that is the same on both sensors there is a 99.99% chance that reading is right. If the pilot only gets a reading from one sensor then it is the same as the single sensor plane case, 98.99%.
Overall, on the two sensor plane if the pilot checks the sensors and accepts the result when the two sensors agree on a reading or when only one sensor gives a reading, they will be getting a correct result 99.97% of the time.
