Sorted data outlier identification

• US 8,533,656 B1
• Filed: 03/08/2013
• Issued: 09/10/2013
• Est. Priority Date: 11/28/2007
• Status: Expired due to Fees

• Current Assignee: Marvell International Limited (Marvell Technology Group Limited)
• Original Assignee: Marvell International Limited (Marvell Technology Group Limited)
• Inventors: Nnaji, Bede C
• Primary Examiner(s): LEVIN, NAUM B
• Application Number:
  US13/790,298
• Time in Patent Office: 186 Days
• Field of Search: 716/111, 716/136, 714/724
• US Class Current: 716/136
• CPC Class Codes: G01R 31/3008 Quiescent current [IDDQ] test or leakage current test

Abstract

Systems and methods for performing a signature analysis on sorted data to identify one or more outliers in a dataset. The method includes determining, with a control platform, a percentile range of a sorted dataset, wherein the percentile range includes a first plurality of qualified vector patterns; determining a signature of the percentile range; determining a signature of a second plurality of vector patterns, wherein the second plurality of vector patterns includes a first test vector pattern that is outside of the percentile range; and comparing the signature of the percentile range to the signature of the second plurality of vector patterns. The method further includes, based at least in part on the comparing, identifying the first vector pattern as one of (i) a qualifying vector pattern or (ii) an outlier.

Claims

1. A system comprising:

a testing platform having an interface communicatively coupled to a device interconnect, wherein the device interconnect is configured to be communicatively coupled to a device; and

a control platform communicatively coupled to the interface, wherein the control platform is configured to

determine a percentile range of a dataset that has been sorted,

determine a signature of the percentile range,

determine a signature of a plurality of vector patterns, wherein the plurality of vector patterns includes a first test vector pattern that is outside of the percentile range,

compare the signature of the percentile range to the signature of the plurality of vector patterns, and

based at least in part on comparing the signature of the percentile range to the signature of the plurality of vector patterns, identify the first test vector pattern as one of (i) a qualified vector pattern or (ii) a non-qualified vector pattern, wherein a qualified vector pattern represents a measurement or observation that is deemed to be properly associated with remaining qualified vector patterns of the dataset.

2. The system of claim 1, wherein the control platform is further configured to:

receive a plurality of quiescent current (IDDQ) measurements from the testing platform, wherein the plurality of IDDQ measurements relate to a device under test;

based at least in part on the plurality of IDDQ measurements, acquire the dataset; and

sort the dataset to generate the dataset that has been sorted.

3. The system of claim 2, wherein the control platform is further configured to:

set each vector pattern of the dataset equal to a lesser absolute value of (i) a left IDDQ differential and (ii) a right IDDQ differential.

4. The system of claim 2, wherein the dataset is a univariate dataset.

5. The system of claim 2, wherein the control platform is further configured to:

sort the dataset by order of magnitude.

6. The system of claim 1, wherein the control platform is further configured to:

determine a nearest pattern smallest delta value (NPSD_K); and

based at least in part on the nearest pattern smallest delta constant, identify the first test vector pattern as one of (i) a qualified vector pattern or (ii) a non-qualified vector pattern.

7. The system of claim 1, wherein the control platform is further configured to:

re-determine the percentile range of the dataset that has been sorted to include the first test vector pattern; and

re-determine the signature of the percentile range.

8. The system of claim 7, wherein the control platform is further configured to:

determine a signature of a second plurality of vector patterns, wherein the second plurality of vector patterns includes a second test vector pattern that is outside of the re-determined percentile range;

compare the re-determined signature to the signature of the second plurality of vector patterns; and

based at least in part on comparing the re-determined signature to the signature of the second plurality of vector patterns, identify the second test vector pattern as one of (i) a qualified vector pattern or (ii) a non-qualified vector pattern.

9. The system of claim 1, wherein the control platform is further configured to:

determine a nearest pattern smallest delta value (NPSD_K); and

identify a non-qualified vector pattern based at least in part on the NPSD_K.

10. The system of claim 9, wherein the control platform is further configured to:

determine the NPSD_K based at least in part on the plurality of vector patterns.

11. The system of claim 1, wherein the control platform is further configured to:

generate a dataset by, based at least in part on a plurality of quiescent current (IDDQ) measurements, acquiring a dataset, wherein the IDDQ measurements relate to a device under test; and

sorting the dataset to generate the dataset that has been sorted.

12. The system of claim 11, wherein the control platform is configured to acquire the dataset by setting each vector pattern of the sorted dataset equal to a lesser absolute value of (i) a left IDDQ differential and (ii) a right IDDQ differential.

13. The system of claim 11, wherein the dataset is a univariate dataset.

14. The system of claim 11, wherein the control platform is configured to sort the dataset by order of magnitude.