5G AMF过载控制N2/NAS双通道流量抑制信令流拆解

Level 1 (Minor)   → 拒绝非紧急MO Data的RRC连接
Level 2 (Moderate) → 拒绝非紧急MO Signaling的RRC连接
Level 3 (Severe)   → 仅允许紧急呼叫和MT服务
Level 4 (Critical) → 仅允许紧急呼叫

sequenceDiagram
    participant AMF
    participant gNB1 as gNB-1
    participant gNB2 as gNB-2
    participant UE1 as UE-1
    participant UE2 as UE-2

    Note over AMF: AMF检测到过载<br/>(CPU 92%, UE Context满)

    rect rgb(255, 200, 200)
        Note over AMF,gNB2: 阶段1: 过载通知下发
        AMF->>gNB1: NGAP OVERLOAD START<br/>OverloadResponse = reject non-emergency MO Data<br/>TrafficLoadReductionIndication = 60%
        AMF->>gNB2: NGAP OVERLOAD START<br/>OverloadResponse = reject non-emergency MO Data<br/>TrafficLoadReductionIndication = 60%
        Note over gNB1: gNB-1按照60%比例<br/>拒绝非紧急MO Data的RRC建立
        Note over gNB2: gNB-2按照60%比例<br/>拒绝非紧急MO Data的RRC建立
    end

    rect rgb(255, 255, 200)
        Note over UE1,UE2: 阶段2: 过载期间流量抑制
        UE1->>gNB1: RRC Connection Request (MO Data)
        gNB1-->>UE1: RRC Connection Reject<br/>waitTime = 8s
        Note over gNB1: 随机丢弃60%的<br/>非紧急MO Data请求

        UE2->>gNB1: RRC Connection Request (Emergency)
        gNB1->>AMF: NGAP Initial UE Message<br/>(Emergency Registration)
        AMF->>gNB1: NGAP Initial Context Setup Request
        gNB1->>UE2: RRC Connection Reconfiguration
        Note over UE2: 紧急呼叫正常建立
    end

    rect rgb(200, 255, 200)
        Note over AMF,gNB2: 阶段3: 过载升级（恶化时）
        AMF->>gNB1: NGAP OVERLOAD START<br/>OverloadResponse = emergency sessions only<br/>TrafficLoadReductionIndication = 90%
        AMF->>gNB2: NGAP OVERLOAD START<br/>OverloadResponse = emergency sessions only
        Note over gNB1: 升级为仅允许紧急呼叫
    end

    rect rgb(200, 200, 255)
        Note over AMF,gNB2: 阶段4: 过载恢复
        Note over AMF: AMF负载恢复至安全水位
        AMF->>gNB1: NGAP OVERLOAD STOP
        AMF->>gNB2: NGAP OVERLOAD STOP
        Note over gNB1: 恢复正常操作
        Note over gNB2: 恢复正常操作
    end

OVERLOAD-START ::= SEQUENCE {
    protocolIEs    ProtocolIE-Container {{OverloadStartIEs}},
    ...
}

OverloadStartIEs NGAP-PROTOCOL-IES ::= {
    { ID id-AMFOverloadResponse   CRITICALITY reject  TYPE OverloadResponse   PRESENCE mandatory  } |
    { ID id-TrafficLoadReductionIndication  CRITICALITY ignore  TYPE TrafficLoadReductionIndication PRESENCE optional } |
    { ID id-OverloadStartNSSAIList  CRITICALITY ignore  TYPE OverloadStartNSSAIList  PRESENCE optional  },
    ...
}

graph TD
    A[OVERLOAD START Message] --> B[OverloadResponse IE<br/>Mandatory]
    A --> C[TrafficLoadReductionIndication<br/>Optional: 0-100%]
    A --> D[OverloadStartNSSAIList<br/>Optional: Rel-16+]

    B --> E[Overload Action 1:<br/>reject-non-emergency-mo-data]
    B --> F[Overload Action 2:<br/>reject-non-emergency-mo-signalling]
    B --> G[Overload Action 3:<br/>permit-emergency-sessions-and-mt-access-only]
    B --> H[Overload Action 4:<br/>permit-emergency-sessions-only]

    D --> I[S-NSSAI 1<br/>e.g. eMBB: 010001]
    D --> J[S-NSSAI 2<br/>e.g. URLLC: 020102]

    style A fill:#ff6b6b,color:#fff
    style B fill:#ffd93d,color:#333
    style C fill:#6bcb77,color:#fff
    style D fill:#4d96ff,color:#fff

对每个符合OverloadResponse过滤条件的RRC请求:
  random_value = random(0, 100)
  if random_value < TrafficLoadReductionIndication:
    → RRC Connection Reject (waitTime)
  else:
    → 正常转发Initial UE Message给AMF

stateDiagram-v2
    [*] --> Idle: UE已注册
    Idle --> Attempting: 发起Registration Request
    Attempting --> Rejected: 收到Reg Reject #22 + T3346
    Rejected --> Backoff: 启动T3346定时器<br/>(如30分钟)
    Backoff --> Backoff_Hold: UE收到寻呼/紧急呼叫
    Backoff_Hold --> Backoff: 紧急呼叫结束<br/>T3346继续运行
    Backoff --> Idle: T3346超时<br/>可重新发起请求
    Backoff --> Attempting: T3346超时后重新尝试

sequenceDiagram
    participant UE1 as UE-1 (新接入)
    participant UE2 as UE-2 (已连接)
    participant gNB as gNB
    participant AMF as AMF (过载中)

    Note over AMF: AMF CPU 95%<br/>触发过载控制

    par N2通道抑制
        AMF->>gNB: NGAP OVERLOAD START<br/>Action: reject MO Data<br/>Reduction: 70%
        Note over gNB: gNB开始按70%概率<br/>丢弃MO Data RRC请求
    and NAS通道抑制准备
        Note over AMF: 配置NAS Congestion Control<br/>Reject Cause #22<br/>T3346 = 30 min
    end

    Note over UE1: 新UE尝试接入
    UE1->>gNB: RRC Connection Request (MO Data)
    Note over gNB: random(0,100)=45 < 70<br/>丢弃此请求
    gNB-->>UE1: RRC Connection Reject<br/>waitTime = 16s
    Note over UE1: UE启动waitTime定时器<br/>16s内不重试

    Note over UE2: 已连接UE发起Service Request
    UE2->>gNB: RRC Connection Request (MO Signaling)
    Note over gNB: MO Signaling不在抑制范围<br/>正常转发
    gNB->>AMF: NGAP Initial UE Message<br/>(Service Request)
    Note over AMF: NAS层判断：AMF过载中<br/>拒绝Service Request
    AMF->>gNB: NGAP Downlink NAS Transport<br/>(Service Reject #22 + T3346=30min)
    gNB->>UE2: RRC: DLInformationTransfer<br/>(NAS Service Reject)
    Note over UE2: UE启动T3346=30分钟<br/>30分钟内不发起Service Request

    Note over AMF: AMF负载恢复
    AMF->>gNB: NGAP OVERLOAD STOP
    Note over gNB: 恢复正常

AMF → NRF: NFUpdate Request
  nfStatus = "overloaded"
  load = 95

NRF → SMF (NFDiscovery Response):
  AMF实例1: status=overloaded, load=95
  AMF实例2: status=registered, load=30

SMF优先选择AMF实例2

stateDiagram-v2
    [*] --> Normal: AMF启动
    Normal --> MinorOverload: CPU > 80%<br/>或UE Context > 70%
    MinorOverload --> Normal: CPU < 60%<br/>持续30秒
    MinorOverload --> ModerateOverload: CPU > 90%<br/>或UE Context > 85%
    ModerateOverload --> MinorOverload: CPU < 75%<br/>持续30秒
    ModerateOverload --> SevereOverload: CPU > 95%<br/>或UE Context > 95%
    SevereOverload --> ModerateOverload: CPU < 85%<br/>持续30秒
    SevereOverload --> CriticalOverload: CPU > 99%<br/>或资源耗尽
    CriticalOverload --> SevereOverload: CPU < 90%<br/>持续60秒

    state Normal {
        [*] --> FullService
        FullService: 接受所有类型请求
    }

    state MinorOverload {
        [*] --> RejectMOData
        RejectMOData: NGAP: reject-non-emergency-mo-data<br/>Reduction: 30%<br/>NAS: T3346=5min
    }

    state ModerateOverload {
        [*] --> RejectMOSignaling
        RejectMOSignaling: NGAP: reject MO Signaling<br/>Reduction: 50%<br/>NAS: T3346=15min
    }

    state SevereOverload {
        [*] --> EmergencyAndMTOnly
        EmergencyAndMTOnly: NGAP: emergency+MT only<br/>Reduction: 80%<br/>NAS: T3346=30min
    }

    state CriticalOverload {
        [*] --> EmergencyOnly
        EmergencyOnly: NGAP: emergency only<br/>Reduction: 95%<br/>NAS: T3346=60min
    }