Non-Guaranteed Bit Rate Bearer

Description

A non-GBR bearer is one of the two fundamental bearer types defined in the Evolved Packet System (EPS) QoS architecture, alongside the GBR bearer. In 3GPP TS 23.203, bearers are classified by whether they carry a guaranteed bit rate commitment: a GBR bearer reserves dedicated network resources at the bearer establishment or modification time, whereas a non-GBR bearer does not reserve any specific bit rate. The non-GBR bearer therefore provides best-effort delivery in which the achievable throughput depends on the instantaneous load and scheduling decisions of the radio access network and the core network.

Architecturally, a non-GBR bearer in EPS is associated with a set of QoS parameters that includes QCI (QoS Class Identifier), ARP (Allocation and Retention Priority), and, when applicable, APN-AMBR (Aggregate Maximum Bit Rate) and UE-AMBR. Because no GBR is reserved, other QoS attributes such as GBR, MBR (Maximum Bit Rate), and per-bearer MBR do not apply to non-GBR bearers. Instead, aggregate bit rate limits are enforced at the APN and UE level to prevent any single subscriber from monopolizing shared resources. This design allows operators to slice their capacity dynamically, granting each non-GBR bearer a share of the remaining resources after GBR traffic has been satisfied.

The default EPS bearer is, by definition, always a non-GBR bearer, which means that every UE attached to an EPC network has at least one non-GBR bearer for basic IP connectivity and signaling. Additional dedicated non-GBR bearers can be established for specific services such as web browsing, email, peer-to-peer file sharing, or background downloads. In LTE and 5G, the scheduler in the eNodeB/gNodeB uses the QCI and ARP to prioritize and allocate radio resources among multiple non-GBR bearers, ensuring fairness and avoiding starvation.

Because non-GBR bearers have no guaranteed throughput, they are commonly used for applications that tolerate variable delay and throughput. Examples include HTTP browsing, bulk file transfer, and most over-the-top (OTT) video streaming services that adapt their bit rate. Operators map these services to specific QCIs such as QCI 5 (IMS signalling), QCI 6–9 (various best-effort data), and the newer 5QI values in 5G. In 5G NR, the non-GBR concept is carried over with QoS flows corresponding to non-GBR bearers; the QoS flow is the finest granularity of QoS treatment in the 5G system.

The non-GBR bearer concept is central to the EPS QoS framework because it enables a scalable, load-aware approach to resource allocation. Rather than reserving resources for every service, operators can reserve them only for services that genuinely require a guaranteed rate (e.g., VoLTE, conversational video), while allowing best-effort applications to compete for the rest. This yields higher spectral efficiency and better overall user experience in congested cells, at the cost of variable performance for non-GBR services.

Purpose & Motivation

The non-GBR bearer was created to address a key limitation of earlier bearer models: reserving guaranteed bit rate resources for every traffic flow is wasteful when most applications do not actually require a hard rate guarantee. In pre-EPS architectures such as GPRS and UMTS release 99, the concept of guaranteed versus non-guaranteed bit rate existed in the form of traffic classes (conversational, streaming, interactive, background), but the mapping to resource reservation was less flexible. EPS formalized the distinction with the GBR/non-GBR split, introducing ARP and aggregate bit rate limits as the tools to manage best-effort traffic.

The primary problem solved by non-GBR bearers is efficient resource utilization. By not reserving rate for delay-tolerant traffic, the network can oversubscribe capacity and serve more users simultaneously, while still delivering acceptable performance through scheduling and congestion control. This directly supports the smartphone era, where users expect always-on connectivity and a mix of applications, most of which do not need guaranteed rates.

Another motivation is policy and charging differentiation. non-GBR bearers are associated with QCIs that allow operators to differentiate treatment of, for example, IMS signalling, buffered streaming, and TCP-based applications, while enforcing fairness via APN-AMBR and UE-AMBR. This enables tiered service offerings and QoS-based charging without requiring per-flow resource reservation, making the model both scalable and commercially attractive.