Co-meet with TAC and discuss AI/ML data sharing project
Minutes
Questions:
What else would you expect regarding AI with ONAP? maybe start from our Control Loop mechanisms and then add AI/ML?
Is there any network automation/autonomy use-case specific model that could be enabled by Acumos Market?
What would an OVP3.0 test suite test?
How can we validate that the algorithms actually work in production?
What is the TM Forum Spec for network intelligence ? IG1230 TMF ANP
What is the common platform and where should it be hosted?
What types of labs do we actually need?
How can Operators share data with open source communities to create algorithms or even models?
Anonymous data sharing issue has been solved for the medical industry stripping out HIPPA data, etc. That is a much bigger problem than sharing networking data.
What are the expectations Operators that create their own AI algorithms and models themselves have from open source communities, is it just the generic Data Analytics Framework platform?
What AI/ML testing already exists with the Operators
Ranny Haiby ia aware of some work that has been done:
Congestion prediction and 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.
Sleeper Cell Detection - Predict a cell going to "sleep" and handover a critical UE (e.g. ambulance) to another cell.
Traffic Steering - Improve Quality of Experience (QoE) by steering UE traffic among multiple cells.
TMF has AI & DAta Analytiucs project. An initiative is is producing an AI Model Data Sheet, Anaother iniziative is on Data Governance with a Data Governance White Paper ( IG 1225)
Brian Freeman some work and royalties may be required to make this work
Possible approaches
is to select a single usecase and use that to understand the challenges around sanitizing the data
create a lab that can be used by the operators to test algorithms
sharing models and algorithms only
share "insights", much like security threat analytics might be shared. Depends on use case. Perhaps network resiliency or fraud use cases would apply to insight sharing?
Can any usecases be inferred from the survey at all? Question #8 most operators already have AI efforts in play for all of these items.
POSSIBLE USECASES:
Sleeper cell detection
Congestion prediction and mitigation
Traffic Steering
Action items
Jim Baker create a wiki page for data consolidation