VoNR — Voice over NR (5G-native voice)¶

Learning objectives¶

After reading this page, you will be able to:

• Explain that VoNR is "a method that carries voice over the NR radio, natively in 5G."
• Explain why VoNR presupposes the IMS (IP Multimedia Subsystem).
• Diagram the flow by which a voice call is established over a dedicated IMS PDU Session and a voice QoS Flow.
• Explain the difference in role between 5QI=5 (IMS signaling) and 5QI=1 (conversational voice, GBR).
• Bridge to the mechanism that establishes voice in areas where VoNR is unavailable (EPS Fallback).

Prerequisites¶

• PDU Session (PDU Session establishment) — VoNR begins by first setting up an IMS PDU Session.
• QoS (the 5G QoS model) — an understanding of QoS Flow / QFI / 5QI / GBR・Non-GBR is essential.
• Registration (initial registration) — before using voice, the UE is assumed to be already registered with the 5GC.
• Supporting: PCF / SMF / UPF / AF

What you'll learn in this chapter¶

• Why — why voice needs a special mechanism in 5G.
• Basic Concept — the idea that "voice is also just IP packets, but carried on a dedicated lane."
• What IMS is (column) — the minimum understanding of the separate subsystem that controls voice calls.
• Architecture — the cast of 5GC / IMS / NG-RAN involved in VoNR.
• Procedure — Call Flow — IMS PDU Session establishment → SIP registration → voice-call QoS Flow establishment.
• How the 5QIs are used — signaling (5QI=5) and conversational voice (5QI=1).
• Comparison with EPC / Release differences / Trouble Shooting / FAQ / Practice

Why — why voice needs a special mechanism in 5G¶

The 5G radio (NR) and core (5GC) are, in the first place, a packet-switched mechanism that carries everything as IP packets. That would suggest you can just send voice as "plain IP packets," but it is not that simple.

Voice calls have requirements that are decisively different from web browsing or file downloads.

• Strict on latency — if the other party's voice arrives one second late, the conversation does not hold together.
• Must not cut out — even a few dropped packets make the sound skip.
• Bandwidth should be guaranteed — you want to secure the bandwidth for voice even during congestion.

In other words, voice needs a "guaranteed dedicated lane." In 5G, this lane is provided as a QoS Flow (see QoS chapter), and additionally, IMS is used as the mechanism that controls with whom and when a call starts/ends. The 5G-native voice that is realized through the coordination of these two (QoS foundation = 5GC, call control = IMS) is VoNR (Voice over NR).

Overview¶

VoNR is realized through the coordination of the following three layers, broadly speaking.

1. IMS (call control) — a separate subsystem that controls "who calls whom" via a protocol called SIP. It sits outside the 5GC.
2. 5GC (providing the foundation) — it sets up an IMS PDU Session to communicate with the IMS, and provides a QoS Flow suited to voice.
3. NG-RAN (radio) — it maps the QoS Flow to a radio bearer (DRB) and carries the voice packets over the NR radio segment.

A voice call is established in the order: "first build the road to the IMS (PDU Session), register to the IMS with SIP, and then, when you actually place a call, add a dedicated voice lane (QoS Flow)."

Basic Concept — a beginner-friendly explanation¶

VoNR can be likened to "a courier service with a dedicated lane."

• Ordinary data communication (web, video) is like normal parcels flowing along a congested general road.
• Voice is a "fragile, time-specified, top-priority" parcel. A dedicated lane (QoS Flow) is provided so it is carried without delay and without loss.
• However, you need a dispatch center that decides "when and to whom to deliver." That is the IMS.

Before placing a call, you first prepare a road on which you can reach the dispatch center (IMS) (the IMS PDU Session). Then, the moment it becomes "we are now starting a call," a top-priority dedicated lane (the voice QoS Flow, 5QI=1) is laid down. When the call ends, the lane is removed.

Architecture¶

graph LR
    UE["UE"] -- "Voice (SIP/RTP)" --> IMS["IMS
(P-CSCF=AF)"]
    UE -- "N1(NAS)" --> AMF["AMF"]
    AMF -- "N11" --> SMF["SMF"]
    SMF -- "N4" --> UPF["UPF"]
    IMS -- "N5" --> PCF["PCF"]
    PCF -- "N7" --> SMF
    SMF -- "N2(via AMF)" --> RAN["NG-RAN"]
    UPF -- "N6" --> IMS

How to read the diagram: the UE's voice (SIP/RTP) first flows over the IMS PDU Session (UE⇔UPF⇔N6⇔IMS). When a call starts, the IMS's P-CSCF, as an AF, requests QoS from the PCF over N5, PCF→N7→SMF receives the policy, and SMF→N4→UPF and SMF→N2 (via AMF)→NG-RAN configure the voice QoS Flow. The point is that control (call on/off = IMS) and foundation (the QoS lane = 5GC) are separated.

Network Function¶

Interface¶

Procedure — Call Flow¶

The VoNR voice call is understood in three stages: (A) IMS PDU Session establishment → (B) SIP registration → (C) voice QoS Flow establishment at origination.

sequenceDiagram
    participant UE
    participant RAN as NG-RAN
    participant SMF
    participant UPF
    participant PCF
    participant IMS as IMS(P-CSCF=AF)

    Note over UE,IMS: (A) IMS PDU Session establishment (DNN=ims)
    UE->>SMF: PDU Session Establishment Request (DNN=ims)
    SMF->>UPF: N4 Session Establishment
    SMF-->>UE: PDU Session established (5QI=5 signaling flow)

    Note over UE,IMS: (B) SIP registration (to IMS)
    UE->>IMS: SIP REGISTER (IMS signaling, over the 5QI=5 flow)
    IMS-->>UE: 200 OK

    Note over UE,IMS: (C) Origination → voice QoS Flow establishment
    UE->>IMS: SIP INVITE (origination)
    IMS->>PCF: N5 Npcf_PolicyAuthorization (voice QoS request)
    PCF->>SMF: N7 SM Policy update (5QI=1 GBR)
    SMF->>UPF: N4 QoS Flow addition (QER)
    SMF->>RAN: N2 QoS profile notification
    RAN-->>UE: Voice DRB setup
    UE->>IMS: Call established (RTP voice, over the 5QI=1 flow)

Explanation of the steps:

1. (A) IMS PDU Session establishment — separately from the ordinary internet one, the UE establishes a PDU Session with DNN=ims (the IMS-dedicated data network name; the value is implementation/operator dependent). At this point there is still no voice lane, and a QoS Flow for IMS signaling (5QI=5, Non-GBR) is provided.
2. (B) SIP registration — through this PDU Session, the UE sends a SIP REGISTER to the IMS, registering with the IMS that "I am here at this number." Thereafter, terminating (incoming) calls also become possible.
3. (C) Origination and voice QoS Flow establishment — when the UE places a call (SIP INVITE), the IMS's P-CSCF, as an AF, requests voice QoS from the PCF over N5. The PCF passes a 5QI=1 (conversational voice, GBR) policy to N7→SMF, and the SMF adds the voice QoS Flow to N4→UPF and N2→NG-RAN. Only here is the guaranteed voice lane laid down, and RTP voice flows over it.

Signal Flow¶

How the 5QIs are used (signaling and voice)¶

In VoNR, two kinds of QoS Flow exist over the same IMS PDU Session.

Picture it as, within a single road (PDU Session), a "signaling lane that is always open" and a "voice lane that is laid down only during a call" running side by side.

flowchart LR
    UE["UE"]
    subgraph PDU["One IMS PDU Session (DNN=ims)"]
        direction TB
        SIG["QoS Flow: 5QI=5 (Non-GBR)
IMS signaling (SIP, etc.)
[From establishment (A) onward, continuously]"]
        VOICE["QoS Flow: 5QI=1 (GBR)
Conversational voice (RTP)
[Only during a call (C); guaranteed bandwidth]"]
    end
    IMS["IMS
(P-CSCF=AF)"]

    UE === SIG
    UE === VOICE
    SIG === IMS
    VOICE === IMS

    classDef sig fill:#eef,stroke:#66a;
    classDef voice fill:#fee,stroke:#c33,stroke-width:2px;
    class SIG sig;
    class VOICE voice;

Comparison with EPC¶

• In 4G: voice over LTE is called VoLTE (Voice over LTE), and it likewise presupposes IMS, using dedicated bearers (QCI=1 for conversational voice, QCI=5 for IMS signaling). The structure is call control = IMS, foundation = EPS (EPC dedicated bearers).
• In 5G: call control being IMS is the same, but the foundation changed from EPS bearer → QoS Flow, and the identifier from QCI → 5QI (QCI=1↔5QI=1, QCI=5↔5QI=5 correspond roughly). The philosophy (IMS + guaranteed lane) carries over from VoLTE, and VoNR is its re-implementation on top of the 5G QoS model.

Release differences¶

• Rel-15: the framework of VoNR (IMS voice over NR) as a 5G system was specified. However, in early deployments there were many areas without VoNR, so EPS Fallback was practically essential in real operation.
• Rel-16 onward: expansion of VoNR support. Detailed feature additions are to be confirmed against each TS (not asserted in this chapter).

Trouble Shooting¶

3GPP Specification¶

• 3GPP TS 23.501 §5.7.4 — the standardized 5QI table (5QI=1 conversational voice/GBR, 5QI=5 IMS signaling/Non-GBR, etc.)
• 3GPP TS 23.501 §5.16.3 — provisions on IMS voice support (section number to be confirmed)
• 3GPP TS 23.502 §4.13.6.1 — EPS Fallback procedure (covered in the EPS Fallback chapter)
• 3GPP TS 23.228 — IMS (IP Multimedia Subsystem) architecture (overview only in this chapter; to be confirmed)

FAQ¶

Q. What is the difference between VoNR and VoLTE? A. Both are voice that uses IMS, but the radio and the foundation differ. VoLTE uses the 4G (LTE) radio and EPS dedicated bearers, whereas VoNR uses the 5G (NR) radio and QoS Flows. The philosophy of using IMS for call control is common to both.

Q. Is IMS part of the 5GC? A. No. The IMS is a separate subsystem that controls voice, etc., and sits outside the 5GC. However, as a point of contact, the IMS's P-CSCF behaves as an AF toward the 5GC and requests QoS from the PCF over N5.

Q. Why are the QoS Flows for voice and for signaling separated? A. Because their roles differ. Call control such as SIP (5QI=5, Non-GBR) is always needed but does not require a bandwidth guarantee. The actual voice (5QI=1, GBR) is only during a call, is strict on latency, and requires a bandwidth guarantee. Separating them lets you secure quality without waste.

Q. What happens in areas where VoNR is unavailable? A. Through EPS Fallback, the UE is temporarily moved to 4G (EPS) at voice-call establishment, and the voice is established as VoLTE. For details, see EPS Fallback.

Summary¶

• VoNR is a method that carries voice over the NR radio, natively in 5G, and is realized through the coordination of call control by IMS and the guaranteed lane provided by the 5GC's QoS Flow.
• A voice call is established in the order (A) IMS PDU Session establishment → (B) SIP registration → (C) voice QoS Flow establishment at origination.
• The IMS's P-CSCF, as an AF, requests QoS from the PCF over N5, and PCF→N7→SMF→(N4 UPF / N2 NG-RAN) lay down the voice lane.
• Over a single IMS PDU Session, two kinds of QoS Flow — 5QI=5 (IMS signaling, Non-GBR) and 5QI=1 (conversational voice, GBR) — coexist.
• In areas without VoNR, EPS Fallback establishes the voice.

Practice — comprehension check and exercises¶

Comprehension check¶

Exercises¶

• Draw a diagram of where the Call Flow is symmetric and where it differs between origination (MO) and termination (MT) (hint: for termination, if the UE is CM-IDLE, Paging comes first).
• Summarize the correspondence between VoLTE (4G) and VoNR (5G) in a table, in the pairs QCI↔5QI and EPS bearer↔QoS Flow.

Next Step¶

• EPS Fallback — the mechanism that establishes voice in areas without VoNR.
• Emergency (emergency calls) — the special handling of emergency calls (unregistered UEs, priority).
• QoS (the 5G QoS model) — go back and deepen the details of QoS Flow / 5QI / GBR.