PDN Gateway User plane function

Description

The PGW-U (PDN Gateway User plane function) is the data plane component resulting from the separation of the traditional P-GW (Packet Data Network Gateway) as defined in 3GPP's Control and User Plane Separation (CUPS) architecture starting in Release 14. It is responsible for the actual forwarding and processing of user data packets between the User Equipment (UE) and external Packet Data Networks (PDNs), such as the internet or an IMS network. While the PGW-C handles signaling and intelligence, the PGW-U handles the high-throughput, latency-sensitive data path.

Architecturally, the PGW-U sits on the user plane path between the Serving Gateway User plane function (SGW-U) and the external network. It interfaces with the SGW-U via the S5/S8-U interface (using GTP-U tunnels) and connects to the external PDN via the SGi interface. Its most critical relationship is with its controlling PGW-C, with which it communicates using the Packet Forwarding Control Protocol (PFCP) as defined in TS 29.244. The PGW-C acts as the master, sending control commands, while the PGW-U acts as the slave, executing those commands on the user plane traffic.

Its operation is rule-driven. The PGW-C installs rules in the PGW-U via PFCP sessions. These rules include Packet Detection Rules (PDRs) that identify which user plane packets to act upon based on criteria like tunnel endpoint identifiers (TEIDs), IP addresses, and port numbers. For packets matching a PDR, the PGW-U applies corresponding Forwarding Action Rules (FARs), which dictate actions such as forwarding the packet to a specific tunnel (e.g., towards the SGW-U or the external network), dropping it, or buffering it. It also applies QoS Enforcement Rules (QERs) to mark packets with DiffServ Code Points (DSCP), enforce uplink/downlink bitrate limits, and apply QoS Flow identification.

The PGW-U performs several key user plane functions. It acts as the anchor point for mobility, meaning the UE's IP address is anchored here, providing session continuity as the user moves and the access network connection point (e.g., the SGW-U) changes. It performs packet inspection and enforcement of charging policies by counting packets/bytes per service data flow as instructed by Usage Reporting Rules (URRs), sending usage reports to the PGW-C for charging. It also handles lawful interception by duplicating packets as required. The decoupled nature of the PGW-U allows it to be implemented as a high-performance, potentially hardware-accelerated function that can be deployed in distributed locations, such as at the network edge, to reduce latency for applications like augmented reality or autonomous vehicles.

Purpose & Motivation

The PGW-U was created to solve the scalability and deployment inflexibility inherent in the monolithic P-GW design of pre-Release 14 EPC networks. In the traditional architecture, the user plane and control plane were tightly integrated within a single physical or virtual network appliance. This coupling meant that scaling to handle more user plane data traffic required scaling the entire node, including the control plane resources, which was inefficient and costly. It also forced a centralized deployment model, preventing operators from placing data forwarding functions closer to users to improve performance.

The driving force behind its development was the industry trend towards network virtualization, cloud-native design, and the need to support diverse, low-latency services anticipated with 5G. The CUPS architecture, formalizing the split, was a direct response. By creating a standalone PGW-U, operators gained the ability to scale the data plane independently and massively, using commodity hardware or specialized data plane accelerators. It also enabled distributed deployment, allowing user plane functions to be placed at central offices, base station aggregation sites, or even at the base station itself (as in Mobile Edge Computing), drastically reducing round-trip time for latency-critical applications.

Furthermore, this separation aligned with Software-Defined Networking (SDN) principles, where the control logic (PGW-C) is centralized and programmable, while the forwarding elements (PGW-U) are simple and distributed. This model simplifies network management, enables faster introduction of new services, and reduces operational costs. The PGW-U concept directly evolved into the User Plane Function (UPF) in the 5G Core network, making it a crucial architectural stepping stone. It solved the problems of monolithic scaling, lack of deployment flexibility, and inability to optimize for both control plane signaling efficiency and user plane throughput/low-latency, which were significant limitations for handling the exponential growth and evolving requirements of mobile broadband traffic.