Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 5 Current »

Date

Attendees

Goals

Discussion items

ItemWhoNotes
Rel 6 Integration TestingReshmaMany bugs and issues have been fixed. Thanks to Reshma, Sandeep and others. SDNC and DMaaP
issues have been resolved.
Link to Integration testing page: xx
Outstanding issue:
  1. Netconf mount on Honeycomb is working. Getting 404 error for netconf config message.
  2. Have to test for Deny message for second CL from Policy.
  3. Lack of extra VM for RAN Sim. Can have limited number of RAN Sim nodes.
Rel 7 planning
We have requested to present to ArchComm on 5/26. Major points related to architecture
  1. Dependency on C&PS - C&PS team has consensus that initial implementation in R7 will meet needs of SON use case.
  2. Dependency on ORAN - In order to prevent major code changes, we need pre-standard version of O-1, VES formats, yang models
  3. Policy - Include Defer from CLC (high priority), Separate Drools instance (depends on Policy)
  4. Modeling - Cell lifecycle - Assume cell addition etc is not part of ONAP. We can assume that data on list of cells is available to C&PS.
  5. Incorporating ML-based use case. See last item.


R7 Wiki page
Guilin (R7) - Use Cases . Need to add OOF-SON in this page.
Use the "Template" Use Case Tracking Template
Incorporating ML-based
SON use cases

Thanks to Vijay and Shankar PN who are PTLs or DCAE and OOF who joined the call. 
We discussed high-level guidance about incorporating ML-based SON use cases.
Consensus was that training of models should be done outside ONAP, and likely to be offline.
ONAP use cases should focus on demonstrating how an ML-trained model can be onboarded. 
This can be done as a recommendation model in OOF, and also as a DCAE MS which leverages work  done
in Frankfurt for the Acumos-DCAE Adapter. See link: Acumos DCAE Integration 

Guidance for new use case discussions:
Generation of data and training of ML-model must be done outside ONAP separately.
ONAP SON solution should include: pre-trained model, source (e.g. enhancement to RANSim) of data needed to apply the pre-trained model

Action items

  • Reshma - netconf debugging email
  • Swami - write to Pam re. Policy
  • No labels