Session Management Congestion Control Experience

Description

Session Management Congestion Control Experience (SMCCE) is a framework standardized in 3GPP Release 17 to enhance Quality of Service (QoS) by proactively managing congestion within the 5G Core Network's session management layer. It operates by enabling the Session Management Function (SMF) to detect congestion conditions based on network load indicators, such as resource utilization at the User Plane Function (UPF) or within specific network slices. Upon detection, the SMF can apply various mitigation policies, which may include throttling data rates for non-priority sessions, rejecting new session establishment requests for certain services, or triggering network-initiated session modifications to reallocate resources. The framework also defines mechanisms for the SMF to report congestion events to the Policy Control Function (PCF), allowing for dynamic policy adjustments, and to the Network Data Analytics Function (NWDAF) for broader network analytics and optimization.

The architecture of SMCCE is integrated within the existing 5G Service-Based Architecture (SBA). Key interactions involve the SMF, PCF, UPF, and the Access and Mobility Management Function (AMF). The SMF acts as the central controller, receiving congestion-related information from the UPF via the N4 interface, such as buffer occupancy or packet drop rates. Based on pre-configured policies or dynamic instructions from the PCF, the SMF can enforce congestion actions. These actions are applied per PDU Session or per QoS Flow, allowing granular control. For example, the SMF might instruct the UPF to apply a lower Guaranteed Flow Bit Rate (GFBR) for a specific QoS Flow or to mark packets for Explicit Congestion Notification (ECN).

SMCCE's role is critical in maintaining service-level agreements (SLAs), especially for network slicing and ultra-reliable low-latency communication (URLLC) services. By preventing uncontrolled congestion, it helps avoid cascading failures and ensures that mission-critical applications receive the necessary resources. The framework supports both reactive and proactive congestion management, with the NWDAF potentially providing predictive insights to trigger actions before severe congestion occurs. This makes SMCCE a foundational element for automated network management and experience assurance in 5G Standalone (SA) deployments.

Purpose & Motivation

SMCCE was created to address the challenge of maintaining consistent user experience and meeting SLA guarantees in increasingly complex and dense 5G networks. Previous 3GPP releases had congestion control mechanisms, but they were often localized (e.g., in the RAN or at the transport layer) or not tightly integrated with the core network's session and policy control. This could lead to suboptimal decisions, such as a transport node dropping packets without the SMF being aware, preventing intelligent session-level remediation. The motivation for SMCCE stemmed from the need for a holistic, session-aware approach to congestion that aligns with 5G's network slicing paradigm, where different slices have vastly different performance and reliability requirements.

Historically, congestion was often managed with simple queue management (e.g., tail drop) in routers or via radio resource scheduling. These methods lack the application and subscriber context available in the 5G core. SMCCE introduces this context by leveraging the SMF's knowledge of active PDU Sessions, their associated QoS profiles, and the network slice they belong to. This allows for more intelligent actions, like selectively degrading best-effort traffic while protecting a URLLC session for factory automation. The framework solves the problem of opaque congestion where the core network is unaware of bottlenecks in the user plane, enabling coordinated mitigation that can preserve revenue-generating services and overall network stability during traffic surges or partial failures.

Key Features

• Congestion detection and reporting from UPF to SMF via N4 interface
• Granular congestion mitigation actions per PDU Session or QoS Flow
• Integration with PCF for dynamic policy-based congestion response
• Support for network slicing-aware congestion management policies
• Mechanisms for network-initiated session modification due to congestion
• Analytics reporting to NWDAF for network-wide congestion insight

Evolution Across Releases

Defining Specifications