Content for TR 23.700-68 Word version: 18.1.0
6.3 Solution #3: MT signalling and data handling for UE with long eDRX cycle in RRC_Inactive state and UPF buffering
6.3.1 Description
6.3.2 Procedures
6.3.3 Impacts on services, entities and interfaces
...
...
6.3 Solution #3: MT signalling and data handling for UE with long eDRX cycle in RRC_Inactive state and UPF buffering p. 17
6.3.1 Description p. 17
This solution resolves Key Issue #1 for MT signalling and data handling for UE with long eDRX cycle in RRC_Inactive state.
As specified in TS 23.501, NG-RAN decides the eDRX in RRC_Inactive state based on the eDRX parameters in IDLE mode. NG-RAN also determines the buffer timer which is used to indicate the time needed for SMF/UPF to buffer the MT data. Before transitioning the UE into RRC_Inactive state, NG-RAN sends the buffer indicator and buffer timer to the AMF to enable data buffering in Core Network.
Then when MT data arrives, the SMF/UPF will buffer the MT data and forward the buffered data to NG-RAN to trigger RAN based paging when the buffer timer expires. If the UE moves out of the original NG-RAN, the data is transfer between the original and new NG-RAN nodes as needed.
If there is a MO data when the UE is during the eDRX in RRC_Inactive state, the NG-RAN forwards the MO data to UPF to disable the data buffering and any MT buffered data is sent to RAN. If the UE enters RRC_CONNECTED without MO data then then RAN can inform the AMF that the UE is no longer in RRC_Inactive and that can trigger delivery of buffered MT to RAN or delivery of MT signalling.
The AMF stores an indication of when the UE is in RRC_Inactive with a long eDRX cycle and uses that indication to determine how to handle paging, how to respond to reachability and event exposure requests, enabling the use of HLcom and SMS over NAS with long eDRX cycles in RRC_Inactive.
6.3.2 Procedures p. 18
Figure 6.3.2-1: high level procedure for MT data handling of UE with long eDRX in RRC_Inactive state
Step 1.
The UE initiates the Registration Procedure and the eDRX parameters in CM-IDLE state are negotiated.
Step 2.
The AMF sends the Inactive Assistance Information to the NG-RAN. The eDRX parameters in CM-IDLE state are included as defined in clause 5.3.3.2.5 of TS 23.501.
Step 3.
Before transitioning the UE into RRC_Inactive state, NG-RAN determines the eDRX parameters for UE in RRC_Inactive state.
Step 4.
If the eDRX cycle in RRC_Inactive state is larger than 10.24s, NG-RAN may send the N2 message to inform the AMF that UE is entering RRC_Inactive state to enable CN buffering. The buffer indicator and the buffer timer are included to indicate to the AMF to enable data buffering until the buffer timer expires. The CM state of the UE is not update by the indication from NG-RAN, i.e. the UE remains in CM-CONNECTED.
Step 5.
If the AMF receives the notification from NG-RAN that the UE will enter RRC_Inactive state and need to buffer downlink data in the CN, AMF sends the buffer indication and buffer timer to SMF to enable data buffering. The AMF stores an indication that the UE is in RRC_Inactive state with a long eDRX cycle and is therefore not reachable until buffer timer expires. Optionally, the buffer timer contains an initial buffer timer (i.e. time until the start of the first PTW) and re-buffer timer (i.e. the eDRX cycle length time between the start of subsequent PTWs) for the case when the UE remains in RRC_Inactive.
If the AMF receives a Namf_EventExposure_Subscribe request for UE Reachability or Availability after DDN failure and if the AMF has stored the indication that the UE is in RRC_Inactive with a long eDRX cycle then the AMF treats the request as if the UE was in CM_IDLE and sends any responses or events for when UE can be paged (i.e. buffer timer expires).
For MT signalling the AMF takes into account the indication that the UE is in RRC_Inactive state with a long eDRX cycle and triggers the MT signalling for when UE can be paged (i.e. buffer timer expires).
If the AMF receives a Namf_MT_EnableUEReachability Request while the AMF has an indication that the UE is in RRC_Inactive with a long eDRX cycle, then the Namf_MT_EnableUEReachability Response is sent back to the requesting NF that the UE is not reachable until when UE can be paged (i.e. buffer timer expires).
Step 6.
The SMF may configure UPF with the new Buffering Action Rule and buffer timer to enable downlink data buffering for the UE until the buffer timer expires.
Step 7.
Based on the new BAR and buffer timer, the UPF starts the buffer timer with the initial buffer timer and enables data buffering for the UE in RRC_Inactive state.
Step 8.
The UE may be moved to RRC_Inactive state. This step may occur in parallel with steps 4 - 7. In case the UE was moved to RRC_Inactive state before the UPF was reconfigured, the RAN node may buffer DL data that is forwarded to the RAN node for the duration of the eDRX cycle or discard the data based on RAN configuration.
Step 9.
When MT data arrives at the UPF, the UPF buffers it until the buffer timer expires.
Step 10.
The UPF may forward the buffered data to the NG-RAN when the buffer timer expires, so NG-RAN can trigger paging for the UE in time. The NG-RAN node buffers the data from the UPF during the paging procedure.
Step 11.
The NG-RAN sends the RAN Paging message to the UE.
Step 12.
The UE responds to paging from RAN.
Step 13.
The response to paging may not be in the same RAN node which moved the UE to RRC_Inactive state. If the UE has moved away from of the RAN node that moved the UE to RRC_Inactive state, the UE will initiate the RNAU procedure and the context will be transferred between the original RAN node and the new RAN node.
If there is no Xn interface between the original and the new RAN nodes, the Core Network can be used to transfer the UE context or buffered data from the original RAN node to the new RAN node. For the UE context in the original RAN node, the AMF can help to retrieve the UE context from the original RAN node and send it to the new RAN node in a container. Similarly, an indirect data forwarding path is built from the original RAN node, the anchor UPF and the new RAN node to enable the buffered data to be transferred.
Step 14.
The MT data is sent to UE.
Step 15.
If there is MO data/signalling triggered connection resume between UE and NG-RAN, the UE enters CM-CONNECTED state.
Step 16.
The NG-RAN notifies the AMF that the UE's RRC state has been changed from RRC_Inactive to RRC_CONNECTED state.
Step 17.
The AMF may also initiate the Nsmf_PDUSession_UpdateSMContext service to notify SMF about the state change.
Step 18.
Accordingly, the SMF decides disables the data buffering in the UPF and the UPF sends the any buffered data to RAN.
Step 19.
If the UE initiated connection resume is triggered by UL data traffic, the NG-RAN sends the UL data to the UPF, which triggers the UPF to disable the data buffering and forward any DL traffic.
6.3.3 Impacts on services, entities and interfaces p. 20
RAN:
- Derive the buffer timer and send it to AMF together with the buffer indicator.
- Proactively inform AMF about the RRC states of UE.
- Support CN assisted AS UE context retrieval and data forwarding.
- Store the indication that the UE is in RRC_Inactive with a long eDRX cycle until the buffer timer expires and replies to requests or sends events if relevant requests were received while it was set.
- Support CN assisted AS UE context retrieval and data forwarding.
- Disable data buffering and forward buffered data when buffer timer expires or there is any UL data traffic.
- Restart buffer timer if there is no DL data buffered or new buffer timer from RAN.
- Disable data buffering and forward buffered data when buffer timer expires or there is any UL data traffic.
