Interface / Protocol Overview¶

Learning objectives¶

After reading this chapter, you will be able to:

• Explain what a reference point is and why it is defined as a "point-to-point connection"
• Explain what an SBI (Service-Based Interface) is and why it is designed as "exposing a common API"
• Explain the relationship and difference between reference points and SBIs using an analogy
• Draw a map of which reference point the major 5GC protocols (NAS / NGAP / PFCP / GTP-U / HTTP/2 / SCTP) run on
• Explain the difference between Control Plane (C-Plane) and User Plane (U-Plane) protocols

Prerequisites¶

• 5GC Overview and SBA — The conceptual introduction to SBA (Service-Based Architecture) and SBI. It is assumed here.
• Network Function Overview — What kind of "box" each NF (AMF/SMF/UPF, etc.) is

This page looks at 5GC not from the NF side (the boxes), but from the side of the lines connecting box to box (interfaces) and the protocols that flow over them.

What you'll learn in this chapter¶

• Why — Why we distinguish between the two ideas of reference points and SBIs
• Basic Concept — The analogy of a reference point = a dedicated window, and an SBI = a web service where common APIs are exposed to each other
• What a reference point is / What an SBI is / The difference between them — Organizing the concepts
• Major protocol map — Which protocol runs on which reference point
• Protocol differences between C-Plane and U-Plane — Why control signaling and user data use different technologies
• Comparison with EPC — The shift from point-to-point IFs (S1/S5, etc.) to SBI

Why — Why distinguish reference points from SBIs¶

5GC is a system in which many NFs (Network Functions) work together. To connect NFs, you must decide "which point talks to which point, about what, and how." 3GPP defines this "way of connecting" in two broad styles.

1. Reference point — A point-to-point approach that fixes and names "two parties, A and B," and decides the protocol used over that segment
2. SBI (Service-Based Interface) — A service-oriented approach in which each NF exposes its services as a common API, and any party that needs it makes a call

The links between Control Plane NFs (e.g., AMF ⇔ SMF) are designed mainly with SBI, while the radio and User Plane segments (e.g., (R)AN ⇔ AMF, gNB ⇔ UPF) are designed mainly with reference points plus dedicated protocols. Because the design philosophy differs even for the same "interface," being able to distinguish these two is the foundation for understanding 5GC.

Overview¶

The interfaces of 5GC can be organized as follows.

• Reference points are called by "N numbers" (N1, N2, N3, N4, N6, …), and the two connected points and the protocol are defined in the dictionary (Interface Dictionary)
• Of these, many of the links between Control Plane NFs are realized in practice as SBI (HTTP/2 over TLS). SBIs are called by a service name in the form of N + NF name, such as "Namf" or "Nsmf"
• The radio and User Plane segments use dedicated protocols (NAS / NGAP / GTP-U / PFCP, etc.) rather than SBI
• Protocols are broadly split into those for the C-Plane (control signaling) and those for the U-Plane (user data), with different required properties (reliability vs. low latency and high throughput)

In other words, "N number = reference point (the name of a segment)" and "SBI service name = exposed API (Namf, etc.)" merely look at things from different angles; many C-Plane reference points are SBI on the inside.

Basic Concept — Explanation for beginners¶

Dedicated window (reference point) vs. common API (SBI)¶

Let us use an analogy of communication between departments in a company.

• A reference point is like a dedicated contact window between the Accounting and General Affairs departments. Who talks to whom, and in what format, is decided from the start. Because the counterpart is fixed, the procedure is clear, but every time a new counterpart is added, you have to build a new window.
• An SBI is like each department exposing a company-wide common request system (API), and any department that needs it makes a request from that screen. Accounting exposes an "expense reimbursement service (equivalent to Namf)," which anyone can call in the same manner. Even as counterparts increase, you just call the already-exposed service.

What a reference point is¶

A reference point is a connection point that 3GPP names by fixing two points: "between which NF/node and which NF/node." In 5GC it is mainly expressed as "N + number," such as N1, N2, N3, N4, N6, …. For each reference point, the protocol and transport used are defined in the dictionary.

Here are just a few representative examples (for the full list, see the Interface Dictionary).

• N1: UE ⇔ AMF. NAS (5GMM/5GSM) signaling. Carried by relaying over N2
• N2: (R)AN ⇔ AMF. Control between the RAN and the core via NGAP
• N3: gNB ⇔ UPF. User Plane transport via GTP-U
• N4: SMF ⇔ UPF. Session control via PFCP
• N6: UPF ⇔ DN (Data Network). External network connection

Because a reference point has "the counterpart decided from the start," it suits segments where you want to engineer performance and reliability carefully, such as the radio segment and the User Plane.

What an SBI is¶

An SBI (Service-Based Interface) is a mechanism in which each NF exposes its own function as a service (API), and other NFs call it. It is the core of SBA (Service-Based Architecture); the conceptual introduction was covered in 5GC Overview and SBA. Here we view it from the IF/protocol side.

• Services are named "N + NF name," such as Namf (AMF), Nsmf (SMF), Nudm (UDM)
• They use HTTP/2 over TLS (with TCP underneath) as the common transport (Protocol Dictionary)
• The payload is mainly JSON, called in the same manner as web APIs (REST-like)

The reason N7/N8/N10/N11/N12/N13/N5/N15/N22/N33/N14/N20/N21/N35/N36/N37/N40 are marked "SBI (HTTP/2 / TLS)" in the Interface Dictionary is that these reference points are links between Control Plane NFs realized in practice as SBI. For example, N11 (AMF ⇔ SMF) is realized by the AMF calling the SMF's exposed service (Nsmf).

The difference between reference points and SBIs¶

The two are not mutually exclusive; many C-Plane reference points have "an N-number segment name" while being SBI on the inside. Let us organize the differences.

Major protocol map¶

This is a map of "which protocol runs on which reference point." The values are consistent with the Interface Dictionary and the Protocol Dictionary.

Correspondence of protocol stacks (C-Plane and U-Plane)¶

graph TD
    subgraph CP["Control Plane (e.g., N1/N2)"]
        NAS["NAS (UE⇔AMF)"] --> NGAP["NGAP (N2)"]
        NGAP --> SCTP["SCTP"]
        SCTP --> IPCP["IP"]
    end
    subgraph UP["User Plane (e.g., N3/N9)"]
        USER["User IP packet"] --> GTPU["GTP-U"]
        GTPU --> UDP["UDP"]
        UDP --> IPUP["IP"]
    end

How to read the diagram: the left is the control signaling stack, where NAS is carried over NGAP, with SCTP/IP underneath. The right is the user data stack, where the user's IP packet is encapsulated by GTP-U, with UDP/IP underneath. It is a layered structure where the upper layer determines "what is conveyed" and the lower layer determines "how it is carried."

Contrast of reference-point connections and the SBI bus¶

graph LR
    subgraph P2P["Point-to-point reference points (U-Plane/radio)"]
        RAN["(R)AN"] -- "N2: NGAP/SCTP" --> AMF1["AMF"]
        gNB["gNB"] -- "N3: GTP-U/UDP" --> UPF1["UPF"]
        SMF1["SMF"] -- "N4: PFCP/UDP" --> UPF1
    end
    subgraph SBA["SBI (between Control Plane NFs: HTTP/2 over TLS)"]
        AMF2["AMF"] --- BUS(("SBI bus"))
        SMF2["SMF"] --- BUS
        UDM["UDM"] --- BUS
        AUSF["AUSF"] --- BUS
        PCF["PCF"] --- BUS
    end

How to read the diagram: the top shows "point-to-point reference points with a fixed counterpart," and the bottom shows "an SBI where each NF exposes its services to each other over a common bus (HTTP/2)." You can see that within the same 5GC, the two styles are used selectively according to the nature of the segment.

Comparison with EPC — From point-to-point IFs to SBI¶

4G (EPC) was designed to connect NFs (more precisely, nodes) with point-to-point interfaces. The biggest difference in 5GC is that it made the Control Plane service-based (SBI).

There are two key points. (1) The Control Plane IFs became SBI, making it easier to add NFs and extend functionality. (2) The User Plane continues to use GTP-U, and the arrangement of an application-layer protocol over SCTP for radio-core control (S1-AP → NGAP) is maintained. In other words, 5GC preserves continuity in the U-Plane and radio side while turning the C-Plane into services. The detailed correspondence with EPC is left to the related Level 1 chapters.

3GPP Specification¶

• Overall architecture (definition of reference points, principles of SBA): TS 23.501
• NAS: TS 24.501
• NGAP: TS 38.413 (SCTP port is TS 38.412)
• GTP-U: TS 29.281
• PFCP: TS 29.244
• General principles of SBI (use of HTTP/2, etc.): TS 29.500; SBI protection (TLS): TS 33.501

FAQ¶

Q. Are "N2" and "Namf" the same thing? A. No. N2 is a reference point (the name of the (R)AN ⇔ AMF segment) and uses NGAP/SCTP. Namf is the name of an SBI service that the AMF exposes and is called over HTTP/2 over TLS. N2 is not SBI.

Q. Is everything with an N number an SBI? A. No. The links between Control Plane NFs (N7/N8/N11/N12, etc.) are SBI, but N1/N2/N3/N4/N6 use dedicated protocols rather than SBI, N9 is GTP-U (U-Plane), and N32 is inter-PLMN interconnection between SEPPs (N32-c/N32-f, PRINS), all with a different positioning (Interface Dictionary).

Q. Why does the User Plane not use SBI (HTTP/2)? A. Because the User Plane needs to transport large volumes of packets with low latency and high throughput, and lightweight encapsulation (GTP-U over UDP) is well suited. The requirements differ from reliability-focused control signaling (HTTP/2 over TLS/TCP).

Q. Which reference point's protocol is NAS? A. It is the protocol of N1 (UE ⇔ AMF), but it is carried by relaying over the radio and N2 segments. The logical exchange between UE and AMF is N1, while its physical carriage path is RRC → NGAP (N2).

Q. How much can the values on this page be trusted? A. Protocols, ports, and transport are made consistent with the Interface Dictionary and Protocol Dictionary. Items marked "(to be confirmed)," such as the clause numbers of individual SBI APIs, are not asserted.

Summary¶

• A reference point is "a point-to-point connection named by fixing two points." N1/N2/N3/N4/N6 are representative
• An SBI (Service-Based Interface) is "a mechanism where NFs expose common APIs (Namf/Nsmf, etc.) to each other." The transport is HTTP/2 over TLS
• Many C-Plane reference points are SBI on the inside. However, N1/N2/N3/N4/N6, N9 (GTP-U), and N32 (between SEPPs) are not SBI
• Major protocol map: NAS(N1) / NGAP(N2, SCTP) / GTP-U(N3, N9, UDP) / PFCP(N4, UDP) / HTTP/2(SBI in general)
• The C-Plane prioritizes reliability, while the U-Plane prioritizes low latency and high throughput, so the protocols used differ
• Against EPC's point-to-point IFs, 5GC turned the Control Plane into SBI (maintaining continuity in the U-Plane and radio side)

Practice — Comprehension check and exercises¶

Comprehension check¶

Exercises¶

• Open the Interface Dictionary and try writing down three "reference points that are SBI" and three "reference points that are not SBI" on your own.
• Try drawing, as a protocol stack, the segments (radio → N2) that an N1 NAS message passes through on its way from the UE to the AMF.

Next Step¶

• Interface Dictionary — Look up the connection points, Protocol, Transport, and Spec of all reference points
• Protocol Dictionary — Details of NAS/NGAP/SCTP/HTTP2/TLS/GTP-U/PFCP
• N2 / N3 / N4 — Individual pages for representative reference points
• Network Function Overview — View 5GC from the "box (NF)" side rather than the lines