System and method for processing signals acquired from multiplexed optical sensors

1. A method, comprising:

receiving streaming data in the form of peak readings developed from spectrum data produced by multiplexed optical sensors of one or more optical fibers, the streaming data comprising wavelength and intensity data associated with the sensors;

determining, for a particular fiber, whether a number of the peak readings is the same as, or differs from, an expected number, N, where N corresponds to a total number of sensors of the particular fiber;

correcting anomalous streaming data in response to determining that the number of the peak readings differs from the expected number, N; and

storing nominal wavelength and intensity streaming data and the corrected wavelength and intensity streaming data in a structured data table indexed by fiber ID and sensor ID.

2. The method of claim 1 , wherein if the number of the peak readings is equal to the expected number, N, then the method comprises storing the nominal wavelength and intensity streaming data in the structured data table by sensor ID.

3. The method of claim 1 , wherein correcting the anomalous streaming data in response to determining that the number of the peak readings is greater than the expected number, N, comprises:

sorting the peak readings based on peak height or intensity in descending order;

selecting the first N peak readings of the sorted peak readings; and

storing wavelength and intensity data for the first N peak readings in the structured data table by sensor ID for each fiber.

4. The method of claim 1 , wherein correcting the anomalous streaming data in response to determining that the number of the peak readings is less than the expected number, N, comprises:

for intervals of the streaming data that include a single peak reading, assigning the single peak readings to their corresponding intervals; and

identifying intervals of the streaming data that do not include a single peak reading as unassigned intervals;

wherein each interval defines an expected wavelength range for each of the sensors.

5. The method of claim 4 , further comprising:

grouping consecutive unassigned intervals;

if the number of the remaining unassigned adjacent intervals is the same as the number of unassigned peak readings, assigning the peak readings to the remaining unassigned adjacent intervals by order and storing wavelength and intensity data associated with the assigned peak readings in the structured data table by sensor ID for each fiber;

if the number of the remaining unassigned adjacent intervals is smaller than the number of unassigned peak readings, no peak readings are assigned to the remaining unassigned adjacent intervals; and

identifying abnormal localized sensors as sensors in which no peak readings have been assigned.

6. The method of claim 1 , further comprising:

identifying fibers with abnormal localized sensors using stateless peak processing; and

analyzing and tracking the abnormal localized sensors using stateful dynamic peak processing.

7. The method of claim 6 , wherein:

(Step A) the peak readings of the abnormal localized sensors are processed in batches using an Expectation-Maximization (EM) algorithm to assign to N expected number of sensor groups, wherein an output of the EM algorithm comprises a mean (μ), standard deviation (σ), and weight (ω) of each sensor group; and

(Step B) if there is any weight smaller than ½N, then N is adjusted to N=N−1 and Step A is repeated.

8. The method of claim 7 , wherein:

(Step C) if no weights are smaller than ½N, the peak readings are grouped into N sensor groups without a sensor group ID; and

(Step D) the method further comprises extracting, for each batch, one set of N−1 decision boundaries, wherein the decision boundaries are used to group the N sensors into N groups.

9. The method of claim 8 , wherein:

the EM algorithm is applied as a moving window to group the decision boundaries between multiple batches; and

the peak readings are assigned to a correct batch with a correct sensor group ID using the decision boundaries.

10. A system, comprising:

a sensor network comprising a network of multiplexed optical sensors of one or more optical fibers;

a processor operatively coupled to the sensor network and configured to:

receive streaming data in the form of peak readings developed from spectrum data produced by the sensors, the streaming data comprising wavelength and intensity data associated with the sensors;

determine, for a particular fiber, whether a number of the peak readings is the same as, or differs from, an expected number, N, where N corresponds to a total number of sensors of the particular fiber; and

correct anomalous streaming data in response to determining that the number of the peak readings differs from the expected number, N; and

a memory operatively coupled to the processor, the memory configured to store nominal wavelength and intensity streaming data and the corrected wavelength and intensity streaming data in a structured data table indexed by fiber ID and sensor ID.

11. The system of claim 10 , wherein if the number of the peak readings is equal to the expected number, N, then the processor stores the nominal wavelength and intensity streaming data in the structured data table by sensor ID.

12. The system of claim 10 , wherein the processor is configured to correct the anomalous streaming data in response to determining that the number of the peak readings is greater than the expected number, N, by:

sorting the peak readings based on peak height or intensity in descending order;

selecting the first N peak readings of the sorted peak readings; and

storing wavelength and intensity data for the first N peak readings in the structured data table by sensor ID for each fiber.

13. The system of claim 10 , wherein the processor is configured to correct the anomalous streaming data in response to determining that the number of the peak readings is less than the expected number, N, by:

for intervals of the streaming data that include a single peak reading, assigning the single peak readings to their corresponding intervals; and

identifying intervals of the streaming data that do not include a single peak reading as unassigned intervals;

wherein each interval defines an expected wavelength range for each of the sensors.

14. The system of claim 13 , the processor is further configured to:

group consecutive unassigned intervals;

if the number of the remaining unassigned adjacent intervals is the same as the number of unassigned peak readings, assign the peak readings to the remaining unassigned adjacent intervals by order and storing wavelength and intensity data associated with the assigned peak readings in the structured data table by sensor ID for each fiber;

if the number of the remaining unassigned adjacent intervals is smaller than the number of unassigned peak readings, no peak readings are assigned by the processor to the remaining unassigned adjacent intervals; and

identify abnormal localized sensors as sensors in which no peak readings have been assigned.

15. The system of claim 10 , wherein the processor is configured to:

identify fibers with abnormal localized sensors using stateless peak processing; and

analyze and track the abnormal localized sensors using stateful dynamic peak processing.

16. The system of claim 15 , wherein:

(Step A) the peak readings of the abnormal localized sensors are processed by the processor in batches using an Expectation-Maximization (EM) algorithm to assign to N expected number of sensor groups, wherein an output of the EM algorithm comprises a mean (μ), standard deviation (σ), and weight (ω) of each sensor group; and

(Step B) if there is any weight smaller than ½N, then N is adjusted by the processor to N=N−1 and Step A is repeated.

17. The system of claim 16 , wherein:

(Step C) if no weights are smaller than ½N, the peak readings are grouped by the processor into N sensor groups without a sensor group ID; and

(Step D) the processor is configured to extract, for each batch, one set of N−1 decision boundaries, wherein the decision boundaries are used to group the N sensors into N groups by the processor.

18. The system of claim 17 , wherein:

the processor is configured to apply the EM algorithm as a moving window to group the decision boundaries between multiple batches; and

the peak readings are assigned by the processor to a correct batch with a correct sensor group ID using the decision boundaries.