Probabilistic diagnosis, in particular for embedded and remote applications

• US 6,691,249 B1
• Filed: 03/22/2000
• Issued: 02/10/2004
• Est. Priority Date: 03/22/2000
• Status: Expired due to Fees

• Current Assignee: Agilent Technologies, Inc.
• Original Assignee: Agilent Technologies, Inc.
• Inventors: Smith, David R., Barford, Lee A.
• Primary Examiner(s): Iqbal, Nadeem
• Assistant Examiner(s): Bonura, Timothy M
• Application Number:
  US09/533,115
• Time in Patent Office: 1,420 Days
• Field of Search: 714/25, 714/26, 714/37, 714/46, 714/57
• US Class Current: 714/25
• CPC Class Codes: G01R 31/3181 Functional testing G01R31/3177 takes precedence
• CPC Class Codes: G06F 11/2257 using expert systems

Abstract

A diagnosis engine for diagnosing a device having a plurality of components receives test results of a set of tests on the device where at least one test has failed, and a model giving the coverage of the tests on the components of the device and information describing probabilistic dependencies between the tests. The diagnosis engine includes a specifier to specify a number N of components that may be simultaneously bad, N being an integer, and a computer to compute, within floating point computation errors, a likelihood that each of subsets of the components having size less than or equal to N are the bad components.

Claims

1. An engine for diagnosing a device having a plurality of components, the engine comprising:

a specifier arranged to specify a number N of components of the plurality of components that are capable of be simultaneously bad, N being a positive integer equal to or greater than 1, and

a computer arranged to compute;

a likelihood that each set of components of the plurality of components having a size less than or equal to N comprise the bad components, based on a model of test coverage of tests on the plurality of components of the device and information as to probabilistic dependencies between the tests;

a probability of a particular set of components being bad and all other components being good, wherein the plurality of components have probabilistically independent states;

a probability of a particular set of shared functions being failed and another particular set of shared functions being passed given that a particular set of components is bad, wherein the states of shared functions, applicable to test the functionality of some of the plurality of components in the same way, are probabilistically independent given component states;

a probability of a particular set of tests being failed and another particular set of shared functions being passed given that a particular set of components is bad and all remaining components of the plurality of components are good, and given that a particular set of shared functions has failed and all remaining shared functions have passed, wherein states of tests applicable on the device are probabilistically independent given component and shared function states; and

test results of a set of tests on the device wherein at least one of the set of tests has failed.

2. The engine of claim 1, further comprising:

an output interface arranged to output, for one or more of the plurality of components, a probability that a component of the plurality of components is bad.

3. The engine of claim 1, further comprising:

another specifier arranged to specify a default value for N components of the plurality of components that are capable of being simultaneously bad.

4. The engine of claim 1, wherein at least one state of each component of the plurality of components is either good or bad, and a state of each shared function is either passed or failed, and the state of each test is either passed or failed.

5. The engine of claim 1, further comprising a memory having a capacity required by the engine during a diagnosis that is less than the memory capacity needed to store the model and the memory capacity needed to store an output.

6. The engine of claim 1, wherein the computer is arranged to compute the failure probability based on the following equation:

7. The engine of claim 6, further comprising:

a memory arranged to store;

an accumulated a sum in accordance with the following equation;

$P<br><br>\left(\mathrm{\pi <br><br>},\mathrm{\phi <br><br>}|C,\stackrel{\_}{C}\right)=\underset{\left(\mathrm{\sigma <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{failed},\stackrel{\_}{\mathrm{\sigma <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{passed}\right)\in <br><br>\mathrm{\kappa <br><br>}<br><br>\left(\mathrm{\Omega <br><br>}\right)}{\sum <br><br>}<br><br>P<br><br>\left(\mathrm{\pi <br><br>},\mathrm{\phi <br><br>}|C,\stackrel{\_}{C},\mathrm{\sigma <br><br>},\stackrel{\_}{\mathrm{\sigma <br><br>}}\right)<br><br>P<br><br>\left(\mathrm{\sigma <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{failed},\stackrel{\_}{\mathrm{\sigma <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{passed}|C,\stackrel{\_}{C}\right)$

accumulated products in accordance with the following equation;

$P<br><br>\left(\mathrm{\pi <br><br>},\mathrm{\phi <br><br>}|C,\stackrel{\_}{C},\mathrm{\sigma <br><br>},\stackrel{\_}{\mathrm{\sigma <br><br>}}\right)=\underset{t\in <br><br>\mathrm{\pi <br><br>}}{\prod <br><br>}<br><br>P<br><br>\left(t<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{passed}|C,\stackrel{\_}{C},\mathrm{\sigma <br><br>},\stackrel{\_}{\mathrm{\sigma <br><br>}}\right)<br><br>\underset{t\in <br><br>\mathrm{\phi <br><br>}}{\prod <br><br>}<br><br>P<br><br>\left(t<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{failed}|C,\stackrel{\_}{C},\mathrm{\sigma <br><br>},\stackrel{\_}{\mathrm{\sigma <br><br>}}\right)$

accumulated products in accordance with the following equation;

$P<br><br>\left(\mathrm{\sigma <br><br>}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{fail}|C,\stackrel{\_}{C}\right)=\underset{s\in <br><br>\mathrm{\sigma <br><br>}}{\prod <br><br>}<br><br>\left(1-\underset{c\in <br><br>C}{\prod <br><br>}<br><br>\left(1-\mathrm{sfcov}<br><br>\left(s,c\right)\right)\right)$

and accumulated products in accordance with the following equation;

$P<br><br>\left(\stackrel{\_}{\mathrm{\sigma <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>\mathrm{pass}|C,\stackrel{\_}{C}\right)=\underset{s\in <br><br>\stackrel{\_}{\mathrm{\sigma <br><br>}}}{\prod <br><br>}<br><br>\underset{c\in <br><br>C}{\prod <br><br>}<br><br>\left(1-\mathrm{sfcov}<br><br>\left(s,c\right)\right).$

8. The engine of claim 1, wherein the computer is arranged to further compute shared functions, the shared functions including Provisional Active Shared Functions and Active Shared Functions.

9. The engine of claim 8, wherein the computer is arranged to compute the shared functions in accordance with the following equation:

10. An engine for diagnosing a device having a plurality of components, the engine comprising:

a first specifier arranged to specify component a priori probabilities, coverages, and shared function coverages;

a second specifier arranged to specify which tests on the components of the device have passed or failed or which tests on the components of the device were not performed;

a third specifier arranged to specify a number N of components of the plurality of components that are capable of being simultaneously bad, N being a positive integer equal to or greater than 1, and

a computer arranged to compute;

a likelihood that each set of the plurality of components of a size less than or equal to N comprise the bad components of the plurality of components;

a probability of a particular set of components being bad and all remaining components of the plurality of components being good, wherein the plurality of components have probabilistically independent states;

a probability of a particular set of shared functions being failed and another particular set of shared functions being passed given that a particular set of components is bad, wherein states of shared functions, applicable to test the functionality of some of the plurality of components in the same way, are probabilistically independent given component states; and

a probability of a particular set of tests being failed and another particular set of shared functions being passed given that a particular set of components is bad and all remaining components are good, and given that a particular set of shared functions has failed and all remaining shared functions have passed, wherein states of tests applicable on the device are probabilistically independent given component and shared function states.

11. A method of diagnosing a device having a plurality of components, the method being performed on the basis of test results of a set of tests on the plurality of components of the device wherein at least one test has failed and the number N of components that are capable of being simultaneously bad, N being a positive integer equal to or greater than 1 and with the aide of a model giving the coverage of tests on the plurality of components of the device and information as to probabilistic dependencies between the tests, the method comprising computing:

a likelihood that each set of the plurality of components having a size less than or equal to N comprise the bad components of said plurality of components;

a probability of any particular set of components of the plurality of components being bad and all remaining components of the plurality of components being good, wherein states of the plurality of components are probabilistically independent;

a probability of any particular set of shared functions being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, wherein states of shared functions, applicable to test functionality of some components of said plurality of components in the same way, are probabilistically independent given component states; and

a probability of any particular set of tests being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, and remaining components of said plurality of components are good, and a particular set of shared functions are failed, and remaining sets of shared functions are passed, wherein states of tests applicable on the device are probabilistically independent given component and shared function states.

12. An engine for diagnosing a device having a plurality of components, the engine comprising:

a specifier arranged to specify a number N of components of the plurality of components that are capable of being simultaneously bad, N being a positive integer equal to or greater than 1; and

a computer arranged to compute;

a likelihood that each set of components of the plurality of components having a size less than or equal to N comprise the bad components, based on a model of the test coverage of tests on the plurality of components of the device and information as to probabilistic dependencies between the tests;

a probability of a particular set of components of the plurality of components being bad and all remaining components of the plurality of components being good, wherein the plurality of components have probabilistically independent states;

a probability of a particular set of tests being failed given that a particular set of components of said plurality of components is bad, and all remaining components of the plurality of components are good, wherein the states of tests applicable to the device are probabilistically independent given component states; and

test results of a set of tests on the device wherein at least one test has failed.

13. A computer program product comprising:

a computer usable medium having computer readable program code means embodied in said medium to perform a method of diagnosing a device having a plurality of components, the method being performed on the basis of test results of a set of tests on the plurality of components of the device wherein at least one test has failed and the number N of components that are capable of being simultaneously bad, N being a positive integer equal to or greater than 1 and with the aide of a model giving the coverage of tests on the plurality of components of the device and information as to probabilistic dependencies between the tests, the computer program product having;

a computer readable program code means for receiving test results of a set of tests on the device wherein at least one test has failed;

a computer readable program code means for receiving a model giving the coverage of tests on the plurality of components of the device and information as to probabilistic dependencies between the tests;

a computer readable program code means for computing;

a likelihood that each set of the plurality of components having a size less than or equal to N comprise the bad components of said plurality of components;

a probability of any particular set of components of the plurality of components being bad and all remaining components of the plurality of components being good, wherein states of the plurality of components are probabilistically independent;

a probability of any particular set of shared functions being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, wherein states of shared functions, applicable to test functionality of some components of said plurality of components in the same way, are probabilistically independent given component states; and

a probability of any particular set of tests being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, and remaining components of said plurality of components are good, and a particular set of shared functions are failed, and remaining sets of shared functions are passed, wherein states of tests applicable on the device are probabilistically independent given component and shared function states.

14. A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform a method of diagnosing a device having a plurality of components, the method being performed on the basis of test results of a set of tests on the plurality of components of the device wherein at least one test has failed and the number N of components that are capable of being simultaneously bad, N being a positive integer equal to or greater than 1 and with the aide of a model giving the coverage of tests on the plurality of components of the device and information as to probabilistic dependencies between the tests, the method comprising computing:

a likelihood that each set of the plurality of components having a size less than or equal to N comprise the bad components of said plurality of components;

a probability of any particular set of components of the plurality of components being bad and all remaining components of the plurality of components being good, wherein states of the plurality of components are probabilistically independent;

a probability of any particular set of shared functions being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, wherein states of shared functions, applicable to test functionality of some components of said plurality of components in the same way, are probabilistically independent given component states; and

a probability of any particular set of tests being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, and remaining components of said plurality of components are good, and a particular set of shared functions are failed, and remaining sets of shared functions are passed, wherein states of tests applicable on the device are probabilistically independent given component and shared function states.

15. An article of manufacture for use in a computer system for performing a method of diagnosing a device having a plurality of components, the method being performed on the basis of test results of a set of tests on the plurality of components of the device wherein at least one test has failed and the number N of components that are capable of being simultaneously bad, N being a positive integer equal to or greater than 1 and with the aide of a model giving the coverage of tests on the plurality of components of the device and information as to probabilistic dependencies between the tests, said article of manufacture comprising a computer usable medium having a computer readable program code means embodied in said medium including:

a computer readable program code means embodied in said computer usable medium for computing;

a likelihood that each set of the plurality of components having a size less than or equal to N comprise the bad components of said plurality of components;

a probability of any particular set of components of the plurality of components being bad and all remaining components of the plurality of components being good, wherein states of the plurality of components are probabilistically independent;

a probability of any particular set of shared functions being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, wherein states of shared functions, applicable to test functionality of some components of said plurality of components in the same way, are probabilistically independent given component states; and

a probability of any particular set of tests being failed and another particular set of shared functions being passed given that a particular set of components of said plurality of components are bad, and remaining components of said plurality of components are good, and a particular set of shared functions are failed, and remaining sets of shared functions are passed, wherein states of tests applicable on the device are probabilistically independent given component and shared function states.