Method and apparatus for performing event-driven diagnostics or prognostics of a network behaviour of a hierarchical optical network

Patent metadata

Field	Value
Publication number	US-10735099-B2
Application number	US-201916267011-A
Country	US
Kind code	B2
Filing date	Feb 4, 2019
Priority date	Feb 7, 2018
Publication date	Aug 4, 2020
Grant date	Aug 4, 2020

Abstract

A method and apparatus for performing event-driven diagnostics and/or prognostics of a network behaviour of a hierarchical optical network comprising the steps of recording at least one set of historical multi-level events representing different hierarchy levels of said optical network; mining of machine learned event patterns within the recorded multi-level events; mapping the determined mined event patterns to a multi-level network topology of said optical network and/or to a channel connectivity of channels through said optical network; and matching observed real-time multi-level target events of said optical network with at least one of the previously determined mined event patterns and performing a unified cause and effect analysis of network states and/or network components of said optical network for a recognized matching event pattern using the network topology and/or channel connectivity associated with the matching event pattern.

First claim

The invention claimed is:

1. A method for performing event-driven diagnostics and/or prognostics of a network behaviour of a hierarchical optical network comprising the steps of:
recording at least one set of historical multi-level events representing different hierarchy levels of said optical network;
mining of machine learned event patterns within the recorded multi-level events;
mapping the determined mined event patterns to a multi-level network topology of said optical network and/or to a channel connectivity of channels through said optical network; and
matching observed real-time multi-level target events of said optical network with at least one of the previously determined mined event patterns and performing a unified cause and effect analysis of network states and/or network components of said optical network for a recognized matching event pattern using the network topology and/or channel connectivity associated with the matching event pattern,
wherein lower-level events are aggregated according to the hierarchy levels of said optical network and the matching is performed on the basis of aggregated high-level events.

2. The method according to claim 1 wherein the unified cause and effect analysis includes a backtracking reactive analysis of observed target events and a forward-looking proactive analysis of future events occurring within the optical network.

3. The method according to claim 1 wherein the recorded multi-level events and/or event patterns are filtered, aggregated and/or sorted.

4. The method according to claim 1 wherein the recorded multi-level events comprise different event types including information events, warning events and/or failure events.

5. The method according to claim 1 wherein the recorded multi-level events comprise events concerning components of said optical network and/or environmental events concerning the environment of said optical network.

6. The method according to claim 1 wherein the recorded multi-level events comprise events from different hierarchy levels of said optical network including system level events, subsystem level events, device level events, component level events and/or events from different protocol layers of a data protocol stack implemented in said optical network.

7. The method according to claim 1 wherein one or more independent sets of historical multi-level events are recorded for different network segments of said optical network and/or
wherein one or more independent sets of historical multi-level events are recorded for the same network segment of said optical network representing different operation time periods of the respective network segment.

8. The method according to claim 1 wherein event patterns for different network segments of said optical network are determined and shared for recognized matching event patterns within the same or other network segments of said optical network belonging to the same or different customers.

9. The method according to claim 1 wherein matching at least one previously determined mined event pattern with the observed real-time multi-level target events of the optical network and the unified cause and effect analysis of the recognized matching event pattern are performed non-intrusively by an event analyzer.

10. The method according to claim 1 wherein the event patterns are mined within the recorded multi-level events according to predefined pattern mining parameter boundaries.

11. The method according to claim 1 wherein if within the sequence of observed real-time multi-level target events of the optical network at least one of the plurality of previously determined mined event patterns is fully or partially recognized as a matching event pattern reactive and/or proactive notifications are issued.

12. The method according to claim 1 wherein event patterns with temporal dependencies are scaled in time based on a geographical system size.

13. The method according to claim 1 wherein for the observed real-time multi-level target events received in a real-time target event stream from the optical network, a calculated similarity is assigned iteratively with respect to the existing previously determined event patterns and a probability that a recognized matching event pattern represents a root cause is calculated.

14. The method according to claim 1 wherein an event database comprising recorded multi-level events is updated with observed real-time multi-level target events.

15. An event analyzer for an optical network,
wherein the event analyzer is adapted to mine machine learned event patterns within recorded sets of multi-level events representing different hierarchy levels and/or protocol layers of said optical network and is adapted to map determined event patterns to a multi-level network topology of said optical network and/or to a channel connectivity of channels provided via said optical network,
wherein the event analyzer is further adapted to match within a sequence of observed real-time multi-level target events of said optical network supplied to said event analyzer at least one of the previously determined mined event patterns and to perform a unified cause and effect analysis of network states and/or network components of said optical network for the recognized matching event pattern using the network topology and/or channel connectivity associated with the matching event pattern,
wherein the event analyzer is adapted to aggregate lower-level events according to the hierarchy levels of said optical network and to perform matching on the basis of aggregated high-level events.

16. An optical network having at least one network segment, wherein each network segment comprises an associated event analyzer according to claim 15 adapted to share event patterns determined by the respective event analyzer with other event analyzers to recognize matching event patterns in the same or different network segments and adapted to report event patterns determined by the respective event analyzer to a central event analyzer of said optical network to recognize a matching event pattern in the optical network.

Assignees

• Adva Optical Networking Se

Inventors

• Rafique Danish
• Eiselt Michael

Classifications

• H04B10/07 Primary

  Arrangements for monitoring or testing transmission systems; Arrangements for fault measurement of transmission systems · CPC title

• H04J2203/0058

  Network management, e.g. Intelligent nets · CPC title

• H04B10/27 Primary

  Arrangements for networking · CPC title

• G06N20/00

  Machine learning · CPC title