ANALYTICS-DRIVEN DYNAMIC NETWORK DESIGN AND CONFIGURATION

• US 20170019302A1
• Filed: 07/13/2015
• Published: 01/19/2017
• Est. Priority Date: 07/13/2015
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Lapiotis, George, Parekh, Chirag, Chang, Henry, Patel, Kiran
• Granted Patent: US 10,439,886 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: H04L 41/0813 characterised by the conditions triggering a change of settings
• CPC Class Codes: H04L 41/122 of virtualised topologies, e.g. software-defined networks [SDN] or network function virtualisation [NFV]
• CPC Class Codes: H04L 41/145 involving simulating, designing, planning or modelling of a network
• CPC Class Codes: H04L 41/16 using machine learning or artificial intelligence
• CPC Class Codes: H04L 43/16 Threshold monitoring

Abstract

A system and method for dynamically (re)configuring a service network based on profile information obtained from a Big Data Analytics platform. Received dynamic situation profiles relative to network states, subscriber states, etc. may be compared against corresponding characteristic situation profiles. If there is a similarity, a dynamic design change action may be effectuated for changing configuration of at least a part of the service network, e.g., a service chaining mechanism, operating to service user data flows of the subscribers.

Claims

1. A network configuration method, comprising:

receiving, from a data analytics platform, a dynamic situation profile with respect to a particular situation defined in terms of one or more service user behavior events and one or more events pertaining to a state of a network servicing a plurality of service users, wherein the dynamic situation profile comprises values corresponding to the service user behavior events and the network state events defining the particular situation;

determining a similarity index between the dynamic situation profile and a characteristic situation profile corresponding to the particular situation, wherein the characteristic situation profile comprises allowable values with respect to the service user behavior events and the network state events defining the particular situation; and

if the similarity index is within an acceptable window, generating one or more control signals to effectuate a dynamic design change action for changing configuration of at least a part of the network operating to service user data flows of the service users.

2. The method as recited in claim 1, wherein the network comprises a service network having a plurality of virtualized service functions and further wherein the design change action comprises at least one of instantiating a new virtualized service function, rearranging one or more existing virtualized service functions, generating one or more instances of an existing virtualized service function, cloning an existing service chain of one or more virtualized service functions, rerouting an existing service chain of one or more virtualized service functions, instantiating additional virtual machine resources, and removing one or more virtualized service functions.

3. The method as recited in claim 1, wherein one or more of the service user behavior events and the network state events are represented as variables comprising at least one of continuous variables, discrete variables, categorical variables, ordinal variables, real variables, Boolean variables, probabilistic variables and deterministic variables.

4. The method as recited in claim 1, wherein the dynamic situation profile and the characteristic situation profile each comprise a vector.

5. The method as recited in claim 1, further comprising:

if the similarity index is not within the acceptable window, appending the dynamic situation profile to a historical profile database and executing a profile learning engine operative to analyze the appended dynamic situation profile in view of a history of profiles with respect to the particular situation; and

receiving a supervisory evaluation, at a design logic editor, with respect to the history of profiles of the particular situation and deriving a candidate characteristic situation profile that is operative as a new characteristic situation profile for the particular situation.

6. The method as recited in claim 5, wherein the supervisory evaluation is received from at least one of a knowledge-based expert system and a human domain expert.

7. The method as recited in claim 2, wherein the service network is configured as a Software-Defined Network (SDN) based on one of OpenFlow protocol and Forwarding and Control Element Separation (ForCES) protocol.

8. The method as recited in claim 1, wherein the similarity index between the dynamic situation profile and the characteristic situation profile is obtained as a vector distance comprising at least one of a Euclidean distance, a Hamming distance, a Chebychev distance, a Minkowsky distance, a Mahalanobis distance and a Bhattacharyya distance.

9. The method as recited in claim 1, wherein the similarity index between the dynamic situation profile and the characteristic situation profile is obtained based on a probabilistic dissimilarity between the two profiles'"'"' events data.

10. The method as recited in claim 1, further comprising:

initializing a set of characteristic situation profiles corresponding to a number of predefined situations; and

updating the set of characteristic situation profiles based on supervisory evaluation of dynamic situation profiles received from the data analytics platform.

11. A network element, comprising:

a profile analyzer operative to receive from a data analytics platform a dynamic situation profile with respect to a particular situation defined in terms of one or more service user behavior events and one or more events pertaining to a state of a network servicing a plurality of service users, wherein the dynamic situation profile comprises values corresponding to the service user behavior events and the network state events defining the particular situation;

a matching module associated with the profile analyzer for determining a similarity index between the dynamic situation profile and a characteristic situation profile corresponding to the particular situation, wherein the characteristic situation profile comprises allowable values with respect to the service user behavior events and the network state events defining the particular situation; and

a dynamic service chain designer, responsive to receiving an indication that the similarity index between the dynamic situation profile and the characteristic situation profile is within an acceptable window, operative for generating one or more control signals to effectuate a dynamic design change action for changing configuration of at least a part of the network operating to service user data flows of the service users.

12. The network element as recited in claim 11, wherein the network comprises a service network having a plurality of virtualized service functions and further wherein the design change action comprises at least one of instantiating a new virtualized service function, rearranging one or more existing virtualized service functions, generating one or more instances of an existing virtualized service function, cloning an existing service chain of one or more virtualized service functions, rerouting an existing service chain of one or more virtualized service functions, instantiating additional virtual machine resources, and removing one or more virtualized service functions.

13. The network element as recited in claim 12, wherein one or more of the service user behavior events and the network state events are represented as variables comprising at least one of continuous variables, discrete variables, categorical variables, ordinal variables, real variables, Boolean variables, probabilistic variables and deterministic variables.

14. The network element as recited in claim 12, wherein the dynamic situation profile and the characteristic situation profile each comprise a vector.

15. The network element as recited in claim 12, further comprising:

a historical profile database that is populated with dynamic situation profiles that do not match corresponding characteristic situation profiles within respective degrees of tolerance;

a profile learning engine operative to generate candidate actionable profiles based on analyzing the historical profiles; and

a design logic editor configured to receive supervisory evaluations with respect to the candidate actionable profiles for providing new characteristic situation profiles.

16. The network element as recited in claim 15, wherein the design logic editor is configured to receive a supervisory evaluation from at least one of a knowledge-based expert system and a human domain expert.

17. The network element as recited in claim 12, wherein the service network is configured as a Software-Defined Network (SDN) based on one of OpenFlow protocol and Forwarding and Control Element Separation (ForCES) protocol.

18. The network element as recited in claim 11, wherein the matching module is configured to generate a similarity index between the dynamic situation profile and the characteristic situation profile that is obtained as a vector distance comprising at least one of a Euclidean distance, a Hamming distance, a Chebychev distance, a Minkowsky distance, a Mahalanobis distance and a Bhattacharyya distance.

19. The network element as recited in claim 11, wherein the matching module is configured to generate a similarity index between the dynamic situation profile and the characteristic situation profile that is obtained based on a probabilistic dissimilarity between the two profiles'"'"' events data.

20. The network element as recited in claim 11, further comprising:

a database including a set of initial characteristic situation profiles corresponding to a number of predefined situations, wherein the set of characteristic situation profiles is updated based on supervisory evaluation of dynamic situation profiles received from the data analytics platform.

21. The network element as recited in claim 11, wherein at least one of the profile analyzer, the matching module and the dynamic service chain designer is implemented as a virtual network function.

22. The network element as recited in claim 11, wherein the profile analyzer, the matching module and the dynamic service chain designer are configured to interoperate with an orchestrator node associated with the network.

23. The network element as recited in claim 11, wherein the profile analyzer, the matching module and the dynamic service chain designer are configured to operate as an orchestrator node associated with the network.

24. A non-transitory computer-readable medium containing instructions stored thereon which, when executed by a computer system configured to operate as an apparatus in a service network, perform the acts:

upon receiving, from a data analytics platform, dynamic situation profiles with respect to one or more service user behavior events and one or more events pertaining to a state of a network servicing a plurality of service users, determining if the dynamic situation profiles match corresponding characteristic situation profiles, wherein the characteristic situation profiles comprises threshold values with respect to the service user behavior events and the network state events;

forwarding one or more dynamic situation profiles that match the corresponding characteristic situation profile to a design action module for generating appropriate control signals for changing configuration of at least a part of the network operating to service user data flows of the service users;

storing one or more dynamic situation profiles that do not match the corresponding characteristic profiles in a historical profile database; and

causing to execute a profile learning module for determining a set of candidate actionable profiles for inclusion into a current characteristic situation profile database under supervisory evaluation.