Local PDN Gateway

Description

The Local PDN Gateway (L-GW) is a core network user plane function that acts as the gateway between the User Equipment (UE) and a local data network or the internet, but is deployed in a decentralized manner, often at the network edge near the radio base station (e.g., eNodeB in LTE, gNB in 5G). Architecturally, it incorporates the functions of a traditional PDN Gateway (PGW), such as IP address allocation, packet filtering, policy enforcement, and charging, but within a localized scope. In LTE, it is a logical component that can be collocated with the HeNB (Home eNodeB) in a residential or enterprise setting for LIPA, or deployed at a network aggregation point for SIPTO. In 5G, its functions are subsumed by the User Plane Function (UPF) when deployed at the edge, with the Local UPF being the equivalent entity.

How it works depends on the deployment scenario. For Local IP Access (LIPA), the L-GW is integrated with the HeNB Gateway or the HeNB itself. When a UE attaches and requests a LIPA connection, the Mobility Management Entity (MME) selects the collocated L-GW, and user plane traffic is routed directly from the HeNB to the L-GW and then to the local network (e.g., a home or enterprise LAN), without traversing the operator's core network. This provides low latency and offloads local traffic. For Selected IP Traffic Offload (SIPTO), the L-GW is deployed at a network aggregation point (like a regional data center). The MME or SGSN selects this L-GW based on traffic routing policies, allowing specific IP traffic (e.g., internet traffic) to exit to the internet locally, reducing load on the central PGW and core network transport.

Key components and interfaces include the S5/S8 interface (GTP-based) for communication with the Serving Gateway (SGW) in 3GPP architectures, though in localized deployments like LIPA, this path may be internal. It also implements the Gi interface toward the external PDN (Packet Data Network). The L-GW performs per-bearer packet processing, applying Quality of Service (QoS) policies, and interacts with the Policy and Charging Rules Function (PCRF) for dynamic policy control. Its role is critical for enabling edge computing, low-latency applications (like industrial IoT, AR/VR), and efficient traffic management by bringing gateway functions closer to the user.

Purpose & Motivation

The L-GW was created to solve the problems of network congestion, latency, and inefficient routing for local traffic. In traditional mobile architectures, all user plane traffic, even traffic destined for a local network in the same building as the base station, was routed through a centralized PGW deep in the core network. This 'tromboning' effect introduced unnecessary latency, consumed backhaul and core network resources, and was inefficient for bandwidth-intensive or delay-sensitive applications.

The initial motivation in Release 10 was to support residential and enterprise femtocell (HeNB) deployments with Local IP Access (LIPA), allowing UEs to access local resources (e.g., printers, media servers) directly without impacting the macro network. This addressed the limitation of previous femtocell architectures that still backhauled all traffic to the core. Subsequently, the concept was extended to Selected IP Traffic Offload (SIPTO) at the network edge to offload internet-bound traffic, reducing load on central gateways and improving user experience for services like video streaming.

Historically, L-GW paved the way for network edge intelligence and traffic localization. It addressed the growing demand for mobile data and the need for more efficient network architectures. The limitations of a purely centralized gateway model became apparent with the rise of IoT and real-time applications. L-GW, and its evolution into the Local UPF in 5G, directly motivated the shift toward distributed user planes and edge computing, which are foundational to 5G's low-latency and high-efficiency promises.

Key Features

• Local IP breakout for direct access to nearby networks (LIPA)
• Selected IP traffic offload (SIPTO) at the network edge
• Collocation with access nodes (e.g., HeNB) for ultra-low latency
• Standard PGW functions (IP allocation, policy enforcement, charging) in a local instance
• Reduces backhaul and core network congestion
• Enables edge computing and low-latency service scenarios

Evolution Across Releases

Defining Specifications