Technique for detecting changes in signals that are measured by quantization
View Patent ↗A technique for detecting changes in a signal that is measured and reported by quantization uses a model that is updated in response to the sampling of quantized values representing the signal. In one stage, (i) frequencies of occurrences of different sampled quantized values in the stage are calculated and (ii) mean frequencies for each of the different sampled quantized values in the stage are calculated and recorded. In a next stage, frequencies of occurrences of different sampled quantized values occurring after an end of the preceding stage are calculated and statistically compared with the mean frequencies of the different sampled quantized values determined in the preceding stage. Dependent on this comparison, a notification may be issued indicating the signal is anomalously changing.
1. A method for determining whether a physical variable in a computer system is anomalously changing, wherein values of the physical variable are reported as quantized values, the method comprising:
calculating a current frequency of occurrences of a particular quantized value of the physical variable in response to an occurrence of the particular quantized value, wherein the physical variable is measured by a physical sensor in the computer system;
performing a Sequential Probability Ratio Test (SPRT) using the current frequency and a reference mean frequency of the particular quantized value to obtain a SPRT value; and
issuing an alarm to a user of the computer system when the SPRT value indicates that the physical variable is anomalously changing.
2. The method of claim 1 , further comprising:
determining the reference mean frequency, comprising:
prior to calculating the current frequency:
sampling a plurality of quantized values of the physical variable,
calculating a mean frequency of occurrences of the particular quantized value in the plurality of quantized values, and
storing the mean frequency as the reference mean frequency.
3. The method of claim 1 , wherein performing the SPRT comprises:
statistically comparing the current frequency to an alternate mean frequency, wherein the alternate mean frequency is a mean frequency of the particular quantized value indicative of the physical variable anomalously changing.
4. The method of claim 1 , wherein performing the SPRT comprises:
calculating a test value, the test value being equal to
SPRT
(
i
-
1
)
+
[
(
x
i
-
M
0
)
2
2
σ
2
-
(
x
i
-
M
1
)
2
2
σ
2
]
,
wherein SPRT(i−1) represents a previously calculated test value, wherein x i represents the current frequency, wherein M 0 represents the reference mean frequency, wherein M 1 represents an alternate mean frequency, and wherein σ 2 represents a variance associated with the reference mean frequency.
5. The method of claim 4 , wherein the alternate mean frequency is equal to the reference mean frequency ± a predetermined value.
6. The method of claim 5 , wherein the predetermined value is a predetermined percentage of the reference mean frequency.
7. The method of claim 4 , wherein the alternate mean frequency is dependent on the variance.
8. The method of claim 4 , further comprising:
comparing the calculated test value to a constant value, wherein the constant value is determined based on a selected error tolerance.
9. The method of claim 8 , further comprising:
calculating the constant value, the constant value being equal to
ln
(
β
1
-
α
)
,
wherein α is a first error tolerance value, and wherein β is a second error tolerance value.
10. The method of claim 8 , further comprising:
calculating the constant value, the constant value being equal to
ln
(
1
-
β
1
-
α
)
,
wherein α is a first error tolerance value, and wherein β is a second error tolerance value.
11. The method of claim 1 , further comprising:
calculating an index that represents how significantly the physical variable is anomalously changing.
12. A method for determining whether a physical variable in a computer system is anomalously changing, comprising:
in a first stage:
sampling a first plurality of quantized values of the physical variable, wherein the physical variable is measured by a physical sensor in the computer system, and
calculating a frequency of occurrences for each different quantized value in the first plurality of quantized values;
in a second stage:
sampling a second plurality of quantized values of the physical variable,
updating the frequency of occurrences based on the second plurality of quantized values, and
calculating a reference mean frequency and a reference variance for each different quantized value in the second plurality of quantized values; and
in a third stage:
sampling a third plurality of quantized values of the physical variable,
calculating a current frequency of occurrences of a particular quantized value selected from the third plurality of quantized values in response to sampling the particular quantized value,
performing a Sequential Probability Ratio Test (SPRT) using the current frequency and the reference mean frequency of the particular quantized value to obtain a SPRT value, and
issuing an alarm to a user of the computer system when the SPRT value indicates that the physical variable is anomalously changing.
13. The method of claim 12 , wherein performing the SPRT comprises:
calculating a test value, the test value being equal to
SPRT
(
i
-
1
)
+
[
(
x
i
-
M
0
)
2
2
σ
2
-
(
x
i
-
M
1
)
2
2
σ
2
]
,
wherein SPRT(i−1) represents a previously calculated test value, wherein x i represents the current frequency, wherein M 0 represents the reference mean frequency, wherein M 1 represents an alternate mean frequency, and wherein σ 2 represents the reference variance.
14. The method of claim 13 , wherein the alternate mean frequency is a mean frequency of the particular quantized value indicative of the physical variable anomalously changing.
15. The method of claim 13 , further comprising:
comparing the calculated test value to a value calculated based on a predetermined error tolerance.
16. A computer-readable medium having recorded therein instructions executable by processing, the instructions to:
sample a first plurality of quantized values representing analog values of a physical variable in a computing system, wherein the physical variable is measured by a physical sensor in the computing system;
record a frequency of occurrences for each different quantized value in the first plurality of quantized values;
calculate a reference mean frequency for each different quantized value in the first plurality of quantized values based on the frequency of occurrences;
sample a second plurality of quantized values representing analog values of the physical variable;
calculate a current frequency of occurrences for each different quantized value in the second plurality of quantized values;
perform a Sequential Probability Ratio Test (SPRT) using the current frequency and the reference mean frequency to obtain a SPRT value; and
issue an alarm to a user of the computing system when the SPRT value indicates that the physical variable is anomalously changing.
17. A computing system, comprising:
a physical sensor arranged to:
monitor a physical variable of the computing system, and
output quantized values representative of actual values of the physical variable; and
an integrated circuit arranged to process instructions for:
recording a reference mean frequency of occurrences of a particular quantized value generated from the physical sensor,
calculating a current frequency of occurrences for the particular quantized value generated from the physical sensor,
performing a Sequential Probability Ratio Test (SPRT) using the current frequency of the particular quantized value and the reference mean frequency of the particular quantized value to obtain a SPRT value, and
issuing an alarm to a user of the computing system when the SPRT value indicates that the physical variable is anomalously changing.
18. The computing system of claim 17 , wherein the physical sensor comprises an analog-to-digital converter.
19. The computing system of claim 17 , wherein instructions for performing the SPRT comprises instructions for:
calculating a test value, the test value being equal to
SPRT
(
i
-
1
)
+
[
(
x
i
-
M
0
)
2
2
σ
2
-
(
x
i
-
M
1
)
2
2
σ
2
]
,
wherein SPRT(i−1) represents a previously calculated test value, wherein x i represents the current frequency, wherein M 0 represents the reference mean frequency, wherein M 1 represents an alternate mean frequency, and wherein σ 2 represents a variance associated with the reference mean frequency.
20. The computing system of claim 17 , wherein the computing system is a server.
21. A technique for determining whether a physical variable in a computer system is anomalously changing, comprising:
sampling a first plurality of quantized values of the physical variable for a first predetermined amount of time, wherein the physical variable is measured by a physical sensor in the computer system;
calculating a frequency of occurrences of a particular quantized value in the first plurality of quantized values;
recording a reference mean frequency of the particular quantized value based on the frequency of occurrences;
sampling a second plurality of quantized values of the physical variable;
calculating a current frequency of occurrences of the particular quantized value in the second plurality of quantized values;
performing a Sequential Probability Ratio Test (SPRT) using the current frequency and the reference mean frequency to obtain a SPRT value; and
issuing an alarm to a user of the computer system when the SPRT value indicates that the physical variable is anomalously changing.
22. The technique of claim 21 , further comprising:
sampling a third plurality of quantized values for a second predetermined amount of time prior to sampling the first plurality of quantized values.
23. The technique of claim 21 , wherein sampling the second plurality of quantized values occurs for a second predetermined amount of time.
24. The method of claim 1 , wherein the physical variable comprises at least one selected from the group consisting of a voltage, a temperature, and a current.
25. The method of claim 12 , wherein the physical variable comprises at least one selected from the group consisting of a voltage, a temperature, and a current.
26. The computer-readable medium of claim 16 , wherein the physical variable comprises at least one selected from the group consisting of a voltage, a temperature, and a current.
27. The computing system of claim 17 , wherein the physical variable comprises at least one selected from the group consisting of a voltage, a temperature, and a current.
28. The technique of claim 21 , wherein the physical variable comprises at least one selected from the group consisting of a voltage, a temperature, and a current.