Granted Patent B2

US 12,744,716 · App. 18/784,197 · Granted Sep 22, 2026

Inventors: Maja Curic (Munich, DE); Sagar Tayal (Ambala, IN); Alecio Pedro Delazari Binotto (Munich, DE); Fernando Luiz Koch (Palm Beach Gardens, FL)

Assignee: International Business Machines Corporation

H04L43/08H04L41/16

Abstract

Claims (70)

1 . A method, in a data processing system, for dynamically implementing reactive actions in edge nodes of a data communication network in response to user equipment (UE) behaviors, the method comprising:

collecting data of UE events in the data communication network to infer UE movements and UE behavior within the data communication network;

executing a machine learning computer model on the collected data of the UE events to predict:

the UE movements,

the UE behavior, and

impacts of the predicted UE movements and the predicted UE behavior on edge node conditions within the data communication network, wherein the impacts of the predicted UE movements and the predicted UE behavior are predicted based on quality of service (QoS) metrics;

evaluating an accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior, wherein the evaluating of the accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior comprises:

grouping a plurality of UEs into UE profiles based on attributes of the plurality of UEs, wherein

UEs of the plurality of UEs having similar attributes are grouped into a UE profile of the UE profiles, and

the attributes comprise at least one of first attributes specifying services in use by the plurality of UEs, second attributes specifying devices utilized by the plurality of UEs, or third attributes specifying session durations of the plurality of UEs; and

predicting an impact of inaccurate predictions on the UE profiles;

executing, based on the accuracy, one or more computer executable rules of a rules based engine, to determine one or more reactive actions, wherein

the one or more reactive actions are for reducing the predicted impact of the inaccurate predictions, and

the reducing of the predicted impact of the inaccurate predictions is based on the QoS metrics; and

sending a recommendation of the one or more reactive actions to a control plane of the data communication network for implementation of at least one of the one or more reactive actions on one or more edge nodes of the edge nodes of the data communication network.

2 . The method of claim 1 , wherein the evaluating of the accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior further comprises identifying incorrect predicted impacts based on actual impacts experienced by the data communication network and one or more UEs affected by the incorrect predicted impacts.

3 . The method of claim 2 , wherein the evaluating of the accuracy of the predicted impacts further comprises identifying a severity of the predicted impacts based on data communication network conditions and priority levels of the one or more UEs affected by the incorrect predicted impacts.

4 . The method of claim 1 , wherein the collected data comprises updates to edge node configurations comprising hardware, software, and network configurations, updates to edge node performance metrics, and UE reports specifying device data of devices corresponding to the plurality of UEs.

5 . The method of claim 4 , wherein the executing of the machine learning computer model on the collected data of the UE events comprises:

extracting features from the updates to edge node configurations, the updates to edge node performance metrics, and the UE reports to generate input features for input to the machine learning computer model; and

processing the input features to identify patterns of input features corresponding to the predicted UE movements, the predicted UE behaviors, and the predicted impacts of the predicted UE movements and the predicted UE behavior on the edge node conditions.

6 . The method of claim 1 , further comprising updating a training of the machine learning computer model based on results of the evaluating of the accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior and the determined one or more reactive actions.

7 . The method of claim 1 , wherein the one or more reactive actions comprise at least one of resource reallocation between nodes of the data communication network, fine-tuning edge node configurations of the edge nodes of the data communication network, or implementing a targeted redundancy for the edge nodes in the data communication network.

8 . The method of claim 1 , further comprising:

selecting, by the control plane, one or more management actions to be executed on the one or more edge nodes of the data communication network based on the recommendation of the one or more reactive actions and one or more established policies for management actions; and

executing the one or more management actions on the one or more edge nodes of the data communication network to modify a configuration of the one or more edge nodes.

9 . The method of claim 1 , wherein the plurality of UEs comprises one or more of a smartphone, laptop computer, vehicle mounted computing device, or mobile computing device, and wherein the data communication network is a wireless mobile network.

10 . A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed in a data processing system, causes the data processing system to dynamically implement reactive actions in edge nodes of a data communication network in response to user equipment (UE) behaviors at least by:

collecting data of UE events in the data communication network to infer UE movements and UE behavior within the data communication network;

executing a machine learning computer model on the collected data of the UE events to predict:

the UE movements,

the UE behavior, and

impacts of the predicted UE movements and the predicted UE behavior on edge node conditions within the data communication network, wherein the impacts of the predicted UE movements and the predicted UE behavior are predicted based on quality of service (QoS) metrics;

evaluating an accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior, wherein the evaluating of the accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior comprises:

grouping a plurality of UEs into UE profiles based on attributes of the plurality of UEs, wherein

UEs of the plurality of UEs having similar attributes are grouped into a UE profile of the UE profiles, and

the attributes comprise at least one of first attributes specifying services in use by the plurality of UEs, second attributes specifying devices utilized by the plurality of UEs, or third attributes specifying session durations of the plurality of UEs; and

predicting an impact of inaccurate predictions on the UE profiles;

executing, based on the accuracy, one or more computer executable rules of a rules based engine, to determine one or more reactive actions, wherein

the one or more reactive actions are for reducing the predicted impact of the inaccurate predictions, and

the reducing of the predicted impact of the inaccurate predictions is based on the QoS metrics; and

sending a recommendation of the one or more reactive actions to a control plane of the data communication network for implementation of at least one of the one or more reactive actions on one or more edge nodes of the edge nodes of the data communication network.

11 . The computer program product of claim 10 , wherein the evaluating of the accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior further comprises identifying incorrect predicted impacts based on actual impacts experienced by the data communication network and one or more UEs affected by the incorrect predicted impacts.

12 . The computer program product of claim 11 , wherein the evaluating of the accuracy of the predicted impacts further comprises identifying a severity of the predicted impacts based on data communication network conditions and priority levels of the one or more UEs affected by the incorrect predicted impacts.

13 . The computer program product of claim 10 , wherein the collected data comprises updates to edge node configurations comprising hardware, software, and network configurations, updates to edge node performance metrics, and UE reports specifying device data of devices corresponding to the plurality of UEs.

14 . The computer program product of claim 13 , wherein the executing of the machine learning computer model on the collected data of the UE events comprises:

extracting features from the updates to edge node configurations, the updates to edge node performance metrics, and the UE reports to generate input features for input to the machine learning computer model; and

processing the input features to identify patterns of input features corresponding to the predicted UE movements, the predicted UE behaviors, and the predicted impacts of the predicted UE movements and the predicted UE behavior on the edge node conditions.

15 . The computer program product of claim 10 , wherein the computer readable program further causes the data processing system to update a training of the machine learning computer model based on results of the evaluating of the accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior, and the determined one or more reactive actions.

16 . The computer program product of claim 10 , wherein the one or more reactive actions comprise at least one of resource reallocation between nodes of the data communication network, fine-tuning edge node configurations of the edge nodes of the data communication network, or implementing a targeted redundancy for the edge nodes in the data communication network.

17 . The computer program product of claim 10 , wherein the computer readable program further causes the data processing system to:

select, by the control plane, one or more management actions to be executed on the one or more edge nodes of the data communication network based on the recommendation of the one or more reactive actions and one or more established policies for management actions; and

execute the one or more management actions on the one or more edge nodes of the data communication network to modify a configuration of the one or more-edge nodes.

18 . An apparatus, comprising:

at least one processor; and

at least one memory coupled to the at least one processor, wherein the at least one memory comprises instructions which, when executed by the at least one processor, cause the at least one processor to dynamically implement reactive actions in edge nodes of a data communication network in response to user equipment (UE) behaviors, at least by:

collecting data of UE events in the data communication network to infer UE movements and UE behavior within the data communication network;

executing a machine learning computer model on the collected data of the UE events to predict:

the UE movements,

the UE behavior, and

impacts of the predicted UE movements and the predicted UE behavior on edge node conditions within the data communication network, wherein the impacts of the predicted UE movements and the predicted UE behavior are predicted based on quality of service (QoS) metrics;

evaluating an accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior, wherein the evaluating of the accuracy of the predicted impacts of the predicted UE movements and the predicted UE behavior comprises:

grouping a plurality of UEs into UE profiles based on attributes of the plurality of UEs, wherein

UEs of the plurality of UEs having similar attributes are grouped into a UE profile of the UE profiles, and

the attributes comprise at least one of first attributes specifying services in use by the plurality of UEs, second attributes specifying devices utilized by the plurality of UEs, or third attributes specifying session durations of the plurality of UEs; and

predicting an impact of inaccurate predictions on the UE profiles;

executing, based on the accuracy, one or more computer executable rules of a rules based engine, to determine one or more reactive actions, wherein

the one or more reactive actions are for reducing the predicted impact of the inaccurate predictions, and

the reducing of the predicted impact of the inaccurate predictions is based on the QoS metrics; and

sending a recommendation of the one or more reactive actions to a control plane of the data communication network for implementation of at least one of the one or more reactive actions on one or more edge nodes of the edge nodes of the data communication network.

Assignments (1)

Continuity (1)

References Cited (11)