Protocol Data Unit Session

Description

The PDU Session is the cornerstone of the 5G Service-Based Architecture (SBA) connectivity model, introduced in 3GPP Release 14 and formalized in TS 23.501. It represents a logical association between a User Equipment (UE) and a Data Network (DN), such as the Internet, an IP Multimedia Subsystem (IMS), or an enterprise network. Unlike the Evolved Packet System (EPS) bearer concept from LTE, the PDU Session provides a much more flexible framework that supports multiple PDU Session types: IPv4, IPv6, IPv4v6, Ethernet, and Unstructured (non-IP). This architectural shift enables 5G to natively support diverse verticals, including industrial IoT, automotive, and fixed wireless access, without requiring complex encapsulation or tunneling overlays.

Architecturally, a PDU Session is established between the UE and the User Plane Function (UPF) via the N1 interface for control signaling and the N3 interface for user data. The control-plane signaling is handled by the Session Management Function (SMF), which communicates with the Access and Mobility Management Function (AMF), the Policy Control Function (PCF), and the UPF over the N4 interface using PFCP (Packet Forwarding Control Protocol). The SMF allocates IP addresses (or prefixes), manages QoS, and selects the appropriate UPF based on factors such as DNN (Data Network Name), S-NSSAI (Single Network Slice Selection Assistance Information), and UE location. The AMF, although not involved in session management per se, provides the transport for NAS Session Management (SM) messages between the UE and SMF.

One of the most significant features of a PDU Session is its support for multiple QoS Flows. Each QoS Flow is identified by a QoS Flow Identifier (QFI) and represents the finest granularity of QoS treatment within a PDU Session. In the downlink, the UPF classifies packets into QoS Flows based on Service Data Flow (SDF) templates provided by the SMF, while in the uplink, the UE uses QoS rules signaled via NAS. This decoupling of the session (connectivity) from the flows (QoS) allows applications with different requirements—such as enhanced Mobile Broadband (eMBB), Ultra-Reliable Low-Latency Communication (URLLC), and massive Machine-Type Communication (mMTC)—to share the same PDU Session while receiving differentiated treatment.

The PDU Session lifecycle is tightly coupled with mobility and network slicing. When a UE moves, the SMF may maintain the session through UPF re-selection or trigger a session release and re-establishment, depending on the session continuity mode (SSC mode 1, 2, or 3). For network slicing, each PDU Session is associated with one S-NSSAI, ensuring that slice-specific resources and policies are applied. In Release 15 and beyond, the PDU Session also supports features such as Ethernet header compression, reflective QoS, and uplink classifiers. The introduction of PDU Set in Release 19 further extends the PDU Session framework to support application-level data units, such as video frames in XR services.

In summary, the PDU Session is not just a connection but a programmable, policy-driven, multi-service conduit that encapsulates 5G's promise of flexibility and service differentiation. It abstracts the complexity of heterogeneous access (3GPP and non-3GPP) and provides a uniform anchor point for mobility, QoS, and charging.

Purpose & Motivation

Prior to 5G, the EPS bearer model in LTE was optimized for IP-based services and lacked native support for non-IP data types such as Ethernet or Unstructured data, which are critical for IoT and industrial applications. The PDU Session was created to address these limitations by providing a more generic and extensible connectivity framework. It decouples the concept of a session from the underlying bearer, allowing multiple QoS flows to coexist and be managed independently. This is essential for supporting the diverse service requirements of 5G, including network slicing, where each slice may require its own session management policies.

The PDU Session also solves the problem of session continuity across different access technologies and mobility scenarios. By defining Session and Service Continuity (SSC) modes, 3GPP enables operators to choose whether to maintain a session when the UE moves (SSC mode 1), break and re-establish immediately (SSC mode 2), or break and re-establish with a new IP address to avoid anchor relocation (SSC mode 3). This flexibility is vital for applications with different tolerance to IP address changes, such as voice over IMS versus HTTP streaming.

Furthermore, the PDU Session architecture supports the separation of control and user plane (CUPS), which allows for scalable and distributed deployments. It also enables edge computing by allowing the SMF to select a UPF close to the UE, reducing latency. The introduction of PDU Sets in Release 19 demonstrates the ongoing evolution of this concept to meet emerging application needs, such as XR and cloud gaming, where the payload of a single application unit may span multiple PDUs and requires coordinated QoS treatment.