[MME] Issue when restarting UE across different TAC's eNodeBs

[s5uishida]

Hi @acetcom

Please let me know one question.

When two SGW-Cs are prepared for each TAC in mme.yaml as below, first the UE connects to EPC via eNodeB(TAC=1) and then the UE stops.

Then when the UE connects to EPC via another eNodeB(TAC=2), MME used the first SGW-C (127.0.0.3) instead of SGW-C (127.0.0.23).

mme.yaml

...
sgwc:
    gtpc:
      - addr: 127.0.0.3
        tac: 1
      - addr: 127.0.0.23
        tac: 2

I would be very glad if this issue was resolved.

Best regards,

--Shigeru

[acetcom]

@s5uishida

The MME selects the SGW when the UE first connects and does not change thereafter. SGW relocation is supported only by X2 handover. Based on TS23.401, I implemented it that way.

Please let me know if I misunderstood anything.

Thanks a lot!
Sukchan

[s5uishida]

Hi @acetcom

Thank you very much for your reply!

Since the UE sends S1AP/NAS-EPS Attach request via the eNodeB with TAC=2, I thought that the previous PDU and PFCP session would be deleted, and then the SGW-C selection would be done again.

[acetcom]

@s5uishida Have you ever seen this scanario in a standard document?

[s5uishida]

Hi @acetcom

According to the following, it says to refer to 3GPP TS 29.303.

3GPP TS 29.244 Annex B.1 CP Selection Function -> B.1.1 General
---
The SGW-C and PGW-C selection function shall follow the principles specified in 3GPP TS 29.303 [25] for the SGW
and PGW selection functions without Control and User Plane Separation.

As the following there was a description of TAU procedure when PGW is already selected.
3GPP TS 29.303 5.2 Procedures for Discovering and Selecting a SGW -> 5.2.1 General

In my case, unlike TAU, I think it's just a case of connecting to a different TAC's eNodeB. I am concerned about the procedure when the UE once goes down and connects again from the eNodeB (TAC=1) to the eNodeB (TAC=2).

[acetcom]

@s5uishida

The SGW change in TAU/Handover are described in standard document. However, in the case of re-attach, there seems to be no mention in the spec. So, I'm just hesitating to implement Open5GS that way.

It seems that this feature will be considered later.

Thanks a lot!
Sukchan

[s5uishida]

Hi @acetcom

Thank you very much for your reply.

I understood.

Also, if I find a specification for this, I'll let you know.

Best regards,

--Shigeru

[github-actions]

This issue has been closed automatically due to lack of activity. This has been done to try and reduce the amount of noise. Please do not comment any further. The Open5GS Team may choose to re-open this issue if necessary.