AI/ML data and model sharing
- Jim Baker
- Beth Cohen
- djhunt
- Ranny Haiby
This is a collective workspace for exploring how to apply open source processes to the development of AI/ML models for use in the operations of intelligent networks.
Ideas on data sharing
Ideas on specific use cases to lead the exploration
If you are an operator or vendor that would like to propose a use case - please add it to the table
If you are an operator or vendor that is interested in one of the listed use cases - please add your name to the table together with proposed contributions, if any
Use Case | Description | Interested Developer | Interested Operator |
|---|---|---|---|
<sample use case> | <In this use case ML is used to predict lightning strikes on cell towers> | Company 1: <Acme Inc.> Contact person1 : <Dr. Emmett Brown> Proposed contribution1:<models, algorithms. etc> Company 2: <Hooli Inc.> Contact person2 : <Gavin Belson> Proposed contribution2:<models, algorithms. etc> | Company 1: <Western Union> Contact person 1: <Marty McFly> Proposed contribution 1:<access to lab, data lake, anonymized data set, etc.> |
Congestion Prediction & Mitigation | This use case will demonstrate how AI/ML may be used to predict congestion and perform closed loop automation for executing configuration changes to mitigate. | Company 1: Samsung Contact person 1: @Ranny Haiby Proposed contribution 1:O-RAN-SC xApp, non-RT RIC, rAPP & AI server Company 2: Contact person2 : Proposed contribution2: | |
Sleeper Cell Detection | Predict a cell going to "sleep" and handover a critical UE (e.g. ambulance) to another cell. | Company 1: Samsung Contact person 1: @Ranny Haiby Proposed contribution 1:O-RAN-SC Non-RT-RIC rApp 2020 October Virtual Technical Event Topic Proposals#2020OctoberVirtualTechnicalEventTopicProposals-ONAP:A1PolicyenforcementwithNon-RTRIC Company 2: Contact person2 : Proposed contribution2: | |
Traffic Steering | Improve Quality of Experience (QoE) by steering UE traffic among multiple cells. | Company 1: Samsung Contact person 1: @Ranny Haiby Proposed contribution 1:O-RAN-SC xApp Company 2: Contact person2 : Proposed contribution2: | |
Soft fault detection and resolution | Detect "soft" faults that are not often caught because they are hidden by the redundant systems. Example, would be faults that bounce for a short time, so are ignored by service assurance. We want to use AI/ML to detect patterns of faults to uncover the ones that might not have an immediate impact on network performance, but will over time as the network degrades. | Company 1: Verizon Contact person 1: @Beth Cohen | |
Deterministic Predictive capacity planning | Ability to detect usage patterns so that the network can be used more efficiently, don't need to built to peak. | Company 1: Verizon Contact person 1: @Beth Cohen | |
Ideas on managing privacy of data and models
One possibility is looking into federated AI learning. For an example, see: https://github.com/IBM/federated-learning-lib
Background data
Results from the EUAG "Intelligent Networks" survey Data_All_210106.pdf
Notes from Feb 17, 2021 EUAG/TAC discussion
Notes from Mar 17, 2021 EUAG/TAC discussion