This page captures the details of the proposed Intelligent Slicing, which is a part of E2E Network Slicing.

Main Contact(s): @LIN MENG, @Swaminathan Seetharaman

Participants: @Chuanyu Chen, @LUKAI,  @Fei Zhang (Ericsson)

Scope and Details

1. Based on following PM data reported from the RAN (Reference: 3GPP TS 28.552 v16.5.0), ONAP acting as NSSMF shall update the maxNbrofConns per cell per S-NSSAI using Machine Learning, to fulfill the maxNbrofConns requirement in Service Profile.

   1. Number of PDU Sessions requested to setup (Section 5.1.1.5.1)

   2. Number of PDU Sessions successfully setup (Section 5.1.1.5.2)

   3. Number of PDU Sessions failed to setup (Section 5.1.1.5.3)

2. The updates shall be sent over O1 interface to the Near-RT RIC. Beyond Guilin, A1 interface shall be used.

3. The training of the ML model shall be done offline.

4. To kick-start the training, fixed rules shall be employed to update the maxNbrofConns based on the above-listed PM data, subject to the limit specified in Service Profile.

5. It is assumed that there is no upper limit to the maxNbrofConns that can be supported by a cell in total, however, the increase in maxNbrofConns shall be moderated.

Supporting Files