Device and method for calibrating a digital sensor
There is provided a calibration device for calibrating a digital sensor ( 3 ), said digital sensor being configured to protest a target digital circuit ( 30 ) fed by a clock signal having a clock period by triggering an alarm depending on a condition between said clock signal and an optimal alarm threshold, said optimal alarm threshold being determined by minimizing a quantity depending on the probability of occurrence of false positives and on the probability of occurrence of false negatives.
1. A calibration device for calibrating a digital sensor, said digital sensor being configured to protect a target digital circuit fed by a clock signal having a clock period by triggering an alarm depending on a triggering condition related to said clock period and to an optimal alarm threshold, said optimal alarm threshold being determined by minimizing a quantity depending on the probability of occurrence of false positives and on the probability of occurrence of false negatives.
2. The calibration device of claim 1 , wherein the probability of false negatives is determined from the probability density of the critical path propagation delay of the target digital circuit and the probability density of the alarm threshold.
3. The calibration device of claim 2 , wherein the probability of false negatives is determined according to the following equation:
FN
=
∫
-
∞
+
∞
p
T
alarm
(
t
alarm
)
∫
t
alarm
+
∞
p
T
c
(
t
c
)
dt
c
dt
alarm
where t alarm designates the alarm threshold, p T alarm designates the probability density of the alarm threshold, t c designates the critical path propagation delay of the target digital circuit, and p T c designates the probability density of the critical path propagation delay of the target digital circuit.
4. The calibration device of claim 1 , wherein the probability of false positives is determined from the probability density of the alarm threshold and the probability density of said clock period of the target digital circuit.
5. The calibration device of claim 4 , wherein the probability of false positives is determined according to the following equation:
FA
=
∫
-
∞
+
∞
p
T
alarm
(
t
alarm
)
∫
-
∞
t
alarm
p
T
clk
(
t
clk
)
dt
clk
dt
alarm
where t alarm designates the alarm threshold, p T alarm designates the probability density of the alarm threshold, p T clk designates the probability density of the clock period and t clk designates the clock period.
6. The calibration device of claim 1 , wherein said quantity depends on the sum of the probability of occurrence of false positives FA weighted by a first weight and the probability of occurrence of false negatives FN weighted by a second weight.
7. The calibration device of claim 1 , wherein said optimal alarm threshold μ alarm is determined according to the following equation:
μ
_
alarm
=
arg
min
μ
alarm
α
FN
+
β
FA
,
in which FA designates the probability of occurrence of false positives, α designates the first weight, FN designates the probability of occurrence of false negatives, and β designates the a second weight.
8. The calibration device of claim 1 , wherein said optimal alarm threshold is determined dynamically during the operation of the digital sensor based on a current operating condition and the critical path propagation delay of the target digital circuit.
9. The calibration device of claim 1 , wherein the digital sensor comprises a sensor critical path comprising a set of elementary gates, said elementary gates being of a unique type, and wherein the number of elementary gates is determined from said optimal alarm threshold and the digital sensor data path.
10. The calibration device of claim 9 , wherein said number of elementary gates n is determined from the optimal alarm threshold μ alarm according to the following equation:
n
_
=
⌊
μ
_
alarm
μ
d
+
1
2
⌋
,
in which μ d designates the digital sensor data path elementary gate mean delay.
11. The calibration device of claim 2 wherein each probability density follows a Gaussian distribution.
12. The calibration device of claim 11 , wherein each probability density comprises a mean value and a variance, wherein if the probability density variances are all equal, and the optimal alarm threshold is determined between the mean value μ T c of the probability density of the critical path propagation delay of the target digital circuit ( 30 ) and the mean value μ T clk of the probability density of the clock period of the target digital circuit.
13. The calibration device of claim 1 , wherein said triggering condition is verified if said optimal alarm threshold exceeds the clock period.
14. A calibration method for calibrating a digital sensor, said digital sensor being configured to protect a target digital circuit fed by a clock signal having a clock period, by triggering an alarm depending on a condition between said clock signal and an optimal alarm threshold, said method comprising determining said optimal alarm threshold by minimizing a quantity depending on the probability of occurrence of false positives and the probability of occurrence of false negatives.
15. A computer program product comprising:
a non-transitory computer readable storage medium; and
instructions stored on the non-transitory computer readable storage medium that, when executed by a processor, cause the processor to calibrate a digital sensor, said digital sensor being configured to protect a target digital circuit fed by a clock signal having a clock period, by triggering an alarm depending on a condition between said clock signal and an optimal alarm threshold, said calibration of the digital sensor comprising determining said optimal alarm threshold by minimizing a quantity depending on the probability of occurrence of false positives and the probability of occurrence of false negatives.