Aggregated Maximum Bit Rate

Description

The Aggregated Maximum Bit Rate (AMBR) is a fundamental Quality of Service (QoS) parameter within the 3GPP Evolved Packet System (EPS) and 5G System (5GS). It operates at two distinct levels: the UE-AMBR (per User Equipment) and the APN-AMBR (per Access Point Name, known as Session-AMBR in 5GS). The UE-AMBR is enforced by the base station (eNodeB in 4G, gNB in 5G) on the radio interface and limits the aggregate bit rate across all non-Guaranteed Bit Rate (non-GBR) bearers or QoS Flows associated with a single UE. The APN-AMBR (or Session-AMBR) is enforced by the core network's User Plane Function (PGW-U/UPF) and applies to the aggregate of all non-GBR bearers/QoS Flows within a specific PDN connection or PDU Session. This hierarchical enforcement ensures control both at the air interface for radio resource fairness and at the Gi/SGi/N6 interface for backhaul and core network resource management.

Architecturally, AMBR is a subscription parameter stored in the Home Subscriber Server (HSS) or Unified Data Management (UDM). During session establishment or modification, this parameter is retrieved and provided to the Policy and Charging Rules Function (PCRF) or Policy Control Function (PCF), which then communicates the authorized AMBR values to the enforcement points via the Packet Data Network Gateway (PGW) or Session Management Function (SMF). The enforcement nodes, such as the eNodeB/gNB and PGW-U/UPF, utilize traffic policing mechanisms like token bucket algorithms to monitor and shape the aggregate traffic, ensuring it does not exceed the configured AMBR limit. Non-compliant packets may be delayed or dropped.

AMBR applies exclusively to non-GBR bearers or QoS Flows, which are used for best-effort services like web browsing, email, and video streaming. This is in contrast to GBR bearers, which have dedicated, guaranteed bandwidth for services like voice or real-time gaming. The separation allows the network to provide strict guarantees for latency-sensitive traffic while efficiently sharing remaining capacity among elastic, non-real-time applications. In 5GS, the concept is enhanced with the Reflective QoS feature, where the UE can derive QoS rules for uplink traffic based on downlink packet markings, but the Session-AMBR remains a network-enforced ceiling.

Its role is pivotal for network efficiency and user experience. By capping the total best-effort bandwidth per user or per session, AMBR prevents network congestion caused by a few heavy users, promotes equitable resource distribution among all connected UEs, and allows operators to implement tiered service plans. It is a key tool for traffic management and monetization, forming the basis for differentiated data packages without requiring per-flow policing for every application.

Purpose & Motivation

AMBR was introduced in 3GPP Release 8 with the EPS to address the limitations of earlier 3G systems in managing the explosive growth of mobile data traffic. Pre-4G networks primarily relied on per-bearer QoS parameters but lacked an efficient mechanism to control the aggregate consumption of a user's multiple simultaneous data sessions (e.g., web, email, app updates). Without AMBR, a user could establish multiple non-GBR bearers, each requesting high bandwidth, and potentially exhaust radio and core network resources, degrading service for other users. AMBR solves this by providing a simple, enforceable ceiling on a user's total best-effort traffic.

The creation of AMBR was motivated by the need for scalable and practical QoS management in all-IP networks. It allows operators to offer 'unlimited' data plans with fair usage policies by technically limiting the maximum achievable throughput, thus protecting network integrity. It also enables service differentiation; for example, a premium subscriber can have a higher UE-AMBR than a basic subscriber, even if they are using the same applications. This addressed a key business requirement for tiered pricing models.

Furthermore, AMBR provides a crucial separation between guaranteed and best-effort service handling. By only applying to non-GBR traffic, it ensures that the network can always reserve necessary resources for GBR services (like VoLTE) to meet their strict latency and reliability requirements, while dynamically sharing the remaining capacity. This architectural principle of aggregate policing for elastic traffic and dedicated resources for inelastic traffic has proven essential for supporting the diverse mix of services on modern mobile networks.