Table of Contents 1. Introduction 3.1 Before S1 Release |
2. S1 Release due to User Inactivity
S1 release may be triggered by either eNB or MME. eNB-triggered release can be caused by:
MME-triggered release can be caused by:
In addition, S1 release can be triggered by the two for other reasons, such as control processing overload, not enough user plane processing resources available, etc.
Figure 1 shows the connections established in user and control planes, and the UE and MME states in the planes, before and after S1 release. Before the release, an EPS bearer and signaling connection are established to support traffic transmission between a user and the network (UE through P-GW). The EPS bearer consists of a DRB, S1 bearer and S5 bearer, while the signaling connection consists of an ECM (RRC + S1 signaling connections), S11 and S5 connections. The UE and MME are in EMM-Registered and ECM-Connected state, while the UE and eNB are in RRC-Connected state.
However, after S1 release, the DRB and downlink S1 bearer are released in the user plane, and the ECM connection (RRC + S1 signaling connections) is lost in the control plane, releasing E-UTRAN resources. It should be noted that at this time only the resources for the downlink S1 bearer are released, and those for the uplink are kept in the network.
The S1 release is different from the one in detach events described in our previous document. In the event of detach, all the resources allocated to a UE by the network are released, and thus the UE transits to EMM-Deregistered state. However, in the event of S1 release, only those allocated by the radio access network (E-UTRAN or eNB) are released, and ones allocated by EPC are kept unreleased. So, the UE remains in EMM-Registered state, transiting to ECM-Idle state. Then later when there is uplink/downlink user traffic, ECM connection and DRB/S1 bearer (downlink) setup is performed, switching the UE state into ECM-Connected, and delivers the traffic.
Figure 1. Connections and States before/after S1 Release
Figure 2 displays the procedures for S1 release triggered by eNB upon detection of user inactivity (Note: Procedures will still be the same even when S1 release is triggered by a cause other than user inactivity). In case S1 release is triggered by MME, Step 1) in Figure 2 will be skipped.
Figure 2. Procedures for S1 Release (eNB-initiated)
1) [eNB → MME] Requesting UE Context Release
The eNB, upon detecting user inactivity, sends the MME a UE Context Release Request message, along with the cause for release, to release the UE context.
2) [MME → S-GW] Requesting S1 Bearer Release
The MME requests the S-GW for release of resources associated with the eNB, a downlink endpoint of the S1 bearer, by sending the S-GW a Release Access Bearers Request message. This way, it informs the S-GW that no downlink traffic can be delivered to the UE.
3) [S-GW] Downlink S1 Bearer Release
The S-GW releases all the downlink S1 bearer resources associated with the UE (eNB related resources, including downlink S1 TEID allocated by eNB, etc.), but keeps the uplink S1 bearer resources (S-GW related resources, including uplink S1 TEID allocated by itself, etc.) unreleased. So, when uplink packets arrive, the eNB can obtain the uplink S1 TEID from the MME, and deliver the packets through the S1 bearer without delay.
4) [MME ← S-GW] Responding to S1 Bearer Release Request
The S-GW acknowledges that the downlink S1 bearer resources have been released by sending the MME a Release Access Bearers Response message. After that, if downlink packets destined to the UE arrive, the S-GW buffers them, and delivers them only after the downlink S1 bearer is re-established. The detailed procedures will be explored in the subsequent document, “EMM Procedure 4. Service Request”.
5) [eNB ← MME] UE Context Release Command
The MME sends the eNB a UE Context Release Command message to release the UE context stored at the eNB.
6) [UE ← eNB] RRC Connection Release
The eNB, upon receiving the command from the MME, deletes all the UE contexts it had. If RRC connection has not been released yet, the eNB sends the UE a RRC Connection Release message to release it. By doing so, the eNB releases all the radio resources and bearers allocated to the UE, and deletes the UE contexts.
7) [eNB → MME] UE Context Release Complete
The eNB sends the MME a UE Context Release Complete message as a response to the request sent in Step 5). The MME then confirms all the UE contexts have been deleted.
8) [MME] S1 Release
The MME deletes all the eNB related information, except for uplink S1 bearer information, in the UE contexts. But it keeps other information not related to the eNB.
In this document, at chapter I. Introduction, p.3 (of the PDF), in the second paragraph it is written:
“From a network’s perspective, S1 release means releasing the S1 signaling and RRC connections in user plane, and the downlink S1 bearer and DRB (Data Radio Bearer) in user plane,”
… I think that the correct should be:
"From a network’s perspective, S1 release means releasing the S1 signaling and RRC connections in CONTROL plane, and the downlink S1 bearer and DRB (Data Radio Bearer) in user plane"
BR
Vassilis
Very good presentation of the information
Very Good and in depth knowldege.
Thanks
Thank you Netmanias! Great Work !
I have a small query. In the shown diagrams "ECGI" value is been removed from PGW and PCRF as well (page 3). Just wanted to understand what message updated it, since there was no message towards PGW in S1 release call flow on page 2.