Knowledge based expert analysis system

• US 6,529,954 B1
• Filed: 06/29/1999
• Issued: 03/04/2003
• Est. Priority Date: 06/29/1999
• Status: Expired due to Term

• Current Assignee: Wandel & Goltermann Technologies, Inc. (Lumentum Holdings Inc.)
• Original Assignee: Wandel & Goltermann Technologies, Inc. (Lumentum Holdings Inc.)
• Inventors: Bruni, Michelle Lee-Farris, Kern, John W. III, Cookmeyer, Eugene N. II
• Primary Examiner(s): Lim, Krisna
• Application Number:
  US09/342,434
• Time in Patent Office: 1,344 Days
• Field of Search: 709/224, 370/241, 370/252, 370/253, 370/254, 714/26, 714/47, 706/47, 706/60
• US Class Current: 709/224
• CPC Class Codes: H04L 41/16 using machine learning or artificial intelligence
• CPC Class Codes: H04L 41/22 comprising specially adapted graphical user interfaces [GUI]
• CPC Class Codes: H04L 41/5009 Determining service level performance parameters or violations of service level contracts, e.g. violations of agreed response time or mean time between failures [MTBF]

Abstract

A rules-based expert analysis system for networks includes a combination of algorithmic and heuristic rules which are grouped in hierarchies in one or more related categories of network performance criteria. Each rule is interdependent in a prioritized manner with one or more other rules, and although the interdependencies are fixed the rule priorities are adaptable based on the user entered objective of the analysis. This allows the rules that are used in the analysis to be defined at run time, instead of fixed rules which are defined at design time and which use fixed threshold values

Claims

1. A method of providing a rules-based expert analysis system for diagnosing network performance by analysis of network performance data, including measured statistics and, or analysis-derived statistics, related to network events and to signal frame transmissions between network addressed stations, the method comprising:

identifying the network performance characteristics associated with each of a plurality of network performance conditions;

defining a plurality of rules, each rule capable of being satisfied by the corresponding presence and, alternately, the absence of an associated one or more of the identified network performance characteristics,

associating each identified network performance condition with the rules which detect the presence and absence of its associated network performance characteristics;

applying the plurality of rules to the network performance data; and

inferring the existence and absence of network performance conditions in response to the satisfaction and non-satisfaction of the rules associated therewith.

2. The method of claim 1, further comprising:

obtaining user observed network performance characteristics prior to analysis; and

prioritizing the rules to be applied to the network performance data in dependence on the observed performance characteristics provided by the user in the step of interviewing.

3. The method of claim 2, wherein said step of prioritizing comprises:

providing each rule with the capability of being programmably enabled or disabled, whereby a rule may not be satisfied unless enabled;

establishing relational interdependencies for the rules which identify the necessary antecedents to the enablement of each rule; and

weighting the interdependency of each rule in dependence on the coincidence and, alternately, the non-coincidence, of the user observed network performance characteristics to the network performance condition associated with the rule.

4. The method of claim 1, wherein one or more of the rules are heuristic based algorithms.

5. A method of providing to a user the expert analysis of network performance data, including, individually and in combination, measured statistics and analysis-derived statistics, related to network events and to signal frame transmissions between network addressed stations, the method comprising:

interviewing the user to determine the purpose of the analysis and the existence of any user known network performance conditions;

identifying a source of network performance data;

obtaining the network performance data;

using a plurality of defined rules to review the network performance data, each rule being capable of being satisfied by the corresponding presence and, alternately, the absence of an associated one or more network performance characteristics which are known to be associated with given network performance conditions;

inferring the existence and absence of network performance conditions in response to the satisfaction and non-satisfaction of the rules associated with the known associated network performance characteristics; and

reporting the results of the review to the user.

6. The method of claim 5, wherein the step of interviewing comprises:

displaying a first user interactive interface which presents the user with a series of statements identifying different network conditions, each statement having accompanying user elected optioned responses which prompt the user in a structured dialog to provide information related to pretest network conditions known to the user; and

prioritizing the rules to be applied to the network performance data in dependence on the user observed network performance characteristics provided in the user elected responses.

7. The method of claim 6, wherein said step of prioritizing comprises:

establishing relational interdependencies for the rules which identify the necessary antecedents to the enablement of each rule; and

altering the interdependency of each rule in dependence on the coincidence and, alternately, the non-coincidence, of the user observed network performance characteristics to the network performance condition associated with the rule.

8. The method of claim 7, wherein said step of altering comprises:

providing each rule with the capability of being programmably enabled or disabled, whereby a rule may not be satisfied unless enabled; and

selecting which rules to be enabled in response to the user elected responses.

9. The method of claim 8, wherein each rule has associated therewith a rule identifier to uniquely identify and differentiate one rule from the other.

10. The method of claim 8, wherein each rule has associated therewith a sampling interval which determines the frequency with which the rule will be evaluated, the sampling interval being adjustable, in seconds, in dependence on the network conditions stated by the user in the interviewing step.

11. The method of claim 8, wherein each rule has associated therewith a priority in relation to one or more other rules, and which indicates the necessary antecedent rules which must first be evaluated prior to the present rule.

12. The method of claim 8, wherein each rule has associated therewith a status, which identifies whether the rule has been satisfied.

13. Memory media, for storing program signals representing an expert analysis algorithm for commanding a computer, in response to user commands, to analyze acquired network performance data, including either or both of measured statistics and analysis-derived statistics, to determine the probable cause of detected network events and signal frame transmissions between network addressed stations, the algorithm comprising:

interviewing the user to determine the purpose of the analysis and the existence of any user known network performance characteristics;

obtaining network performance data;

performing a rules based analysis of the network performance data using a plurality of defined rules, each rule being capable of being satisfied by the corresponding presence and;

alternately, the absence of an associated one or more network performance characteristics which are known to be associated with given network performance conditions;

inferring the existence and absence of network performance conditions in response to the satisfaction and non-satisfaction of the rules associated with the known associated network performance characteristics; and

reporting the inferences regarding network performance conditions to the user.

14. The memory media of claim 13, wherein the step of interviewing comprises:

displaying a first user interactive interface which presents the user with a series of statements identifying different network conditions, each statement having accompanying user elected optioned responses which prompt the user in a structured dialog to provide information related to pretest network conditions known to the user; and

prioritizing the rules to be applied to the network performance data in dependence on the user observed network performance characteristics provided in the user elected responses.

15. The memory media of claim 14, wherein the step of prioritizing comprises:

establishing relational interdependencies for the rules which identify the necessary antecedents to the enablement of each rule; and

altering the interdependency of each rule in dependence on the coincidence and, alternately, the non-coincidence, of the user observed network performance characteristics to the network performance condition associated with the rule.

16. The memory media of claim 15, wherein said step of altering comprises:

providing each rule with the capability of being programmably enabled or disabled, whereby a rule may not be satisfied unless enabled; and

selecting which rules to be enabled in response to the user elected responses.

17. The memory media of claim 16, wherein each rule has associated therewith a rule identifier to uniquely identify and differentiate one rule from the other.

18. The memory media of claim 16, wherein each rule has associated therewith a sampling interval which determines the frequency with which the rule will be evaluated, the sampling interval being adjustable, in seconds, in dependence on the network conditions stated by the user in the interviewing step.

19. The memory media of claim 16, wherein each rule has associated therewith a priority in relation to one or more other rules, and which indicates the necessary antecedent rules which must first be evaluated prior to the present rule.

20. The memory media of claim 16, wherein each rule has associated therewith a status, which identifies whether the rule has been satisfied.