NTN fixed-delay RFsim: UE reaches PBCH/SIB1 but does not progress to visible AMF-side UE activity

[kouamdo]

Summary

We are validating an OAI + Open5GS NTN setup with RFsim. In the fixed-delay NTN baseline, the UE consistently reaches PBCH decode, initial sync, and SIB1 decode, but we do not see successful progression to AMF-visible UE activity.

At the same time, the gNB does not show convincing PRACH / random access progress. This suggests the issue is likely between successful SIB1 acquisition and random access / RRC setup, rather than in Open5GS core-side PDU session or UPF/NAT handling.

Environment

• OAI branch: develop
• OAI build includes nr-softmodem, nr-uesoftmodem, and OAI_SIMU:BOOL=ON
• Open5GS: 2.7.7
• Scenario: NTN fixed-delay baseline over RFsim
• Band: 254
• DL freq: 2488400000
• UL offset: -873500000
• Numerology: 0
• N_RB_DL: 25
• SSB: 60
• PLMN: 999/70

What Works

gNB / core side

• gNB attaches to AMF
• Received NGSetupResponse from AMF is visible

UE side

• pbch decoded sucessfully
• Initial sync successful
• UE synchronized!
• SIB1 decoded

What Does Not Work

• No convincing UE progression into random access / RRC setup
• No AMF-side UE traces such as:
  • InitialUEMessage
  • RAN_UE
  • Registration request
  • Authentication

Relevant UE Log Extract

[PHY]    Initial sync: pbch decoded sucessfully, ssb index 0
[PHY]    Initial sync successful, PCI: 0
[PHY]    UE synchronized!
[NR_RRC] SIB1 decoded

Relevant gNB Log Observation

The gNB reaches NG setup, but repeated PRACH-related indicators remain weak / inconclusive, for example repeated prach_I0 = 0.0 dB, and we do not see a clear successful random access / RRC setup sequence leading to AMF UE visibility.

Current Interpretation

The issue appears to be before the core-network PDU/session plane. In other words:

• this does not currently look like a UPF NAT or DN breakout problem
• it looks more like an issue between NTN sync completion and random access / RRC progression

Question

In the fixed-delay NTN baseline, after successful PBCH and SIB1 decode, are there known constraints or missing runtime/configuration parameters in OAI that can prevent PRACH / RRC progression in this setup?

In particular, is there any known limitation for NTN RFsim fixed-delay mode where UE sync succeeds but random access does not complete reliably?

Additional Note

We also tested LEO staged options separately, but this issue report is specifically about the simpler fixed-delay baseline because it already reproduces the gap between SIB1 decoded and AMF-visible UE activity.

[kouamdo]

This link for better details: https://github.com/kouamdo/load-balance-tn-ntn-5g/blob/main/docs%2Fntn-status-summary.md

[kouamdo]

Thanks for the suggestion.

We reproduced the issue again using the fixed-delay NTN baseline, so that the capture is not mixed with LEO Doppler effects.

We collected the following artifacts:

1. UE and gNB logs on the OAI side
2. AMF log on the Open5GS side
3. tcpdump capture of NGAP/SCTP traffic on loopback (sctp port 38412)

The goal is to verify whether UE-related signaling is actually triggered and sent by the gNB, and whether any InitialUEMessage or NAS-related signaling reaches the AMF.

Our current observation is:

• the gNB successfully attaches to the AMF
• the UE reaches PBCH / initial sync / SIB1
• but no UE-related trace appears in amf.log

We have uploaded the logs and the packet capture to our remote repository and are sharing the links here for analysis.