How it Works

This section describes how this feature works.

N2 Overload Control

When the congestion control feature is enabled, and a congestion threshold is exceeded, the AMF invokes congestion control policies. The AMF informs the control policies to the gNB, which throttles the traffic using the NGAP Overload Start or Stop messages. The AMF sends an NGAP Overload Start message to the gNBs to which it is connected. In the Overload Response IE that the AMF sends to gNBs, the AMF requests the gNBs to reject or allow certain sessions.

After the congestion is cleared, the AMF sends the NGAP Overload Stop message to the NG-RAN node indicating that AMF is resuming regular operations.

NAS Congestion Control

In the overload condition, the AMF rejects the NAS messages from a UE using a 5G-RAN. When AMF rejects a NAS request due to congestion, AMF sends the T3346 IE using the specified T3346 value. With the Mobility Management back-off timer running, the UE can initiate only the Deregistration procedures and procedures that are not affected by the congestion control, such as emergency services and mobile-terminated services.

During a congestion situation, AMF rejects the following requests with the 5GMM cause as Congestion and the T3346 timer value in the Registration Request (including Mobility and Periodic Registration Request) and service requests. The AMF includes the timer value in the Deregistration Request, which UE invokes during the admin clear subscriber process.

Suppose the AMF rejects a Registration Request or service request with the T3346 timer value higher than the sum of the UE's Periodic Registration Update timer T3512 and the Implicit Deregistration timer. In that case, the AMF adjusts the mobile reachable timer and the implicit deregistration timer, or both. With this adjustment, the AMF does not implicitly deregister the UE while the Mobility Management back-off timer is in-progress.

The AMF does not preserve the back-off-timer value which is sent to UE.

For information on the T3346 timer, see the Session Timers chapter.

With the Overload Control feature configured:

  • The protocol-ep pod has X minute timer running locally. On expiry of the timer, the protocol-ep pod identifies the system overload state and accordingly sends the overload start or stop message.

  • On configuration change, AMF sends overload Stop if overload start was sent earlier. Further, AMF continues to monitor the overload state and send stop/start messages accordingly.

  • When the standby Protocol-ep pod becomes active, it collects the system load, determines the overload state, and sends the corresponding message. If AMF identifies an overload situation, the Protocol-ep pod sends an Overload Start message, else, sends the overload stop message if AMF is no longer overloaded.