Radio Access Bearer Sub-flows

Description

RAB sub-flows are a UTRAN-specific concept introduced in Release 4 that allows a single Radio Access Bearer to be realized through one or more sub-flows, each corresponding to a NAS (Non-Access Stratum) service data stream with distinct QoS characteristics. This mechanism enables multiplexing of multiple service data streams with different reliability requirements within a single RAB, providing efficient use of radio resources while maintaining service differentiation.

The concept is defined in 3GPP TS 25.401 (UTRAN overall description) and TS 25.415 (UTRAN Iu interface user plane protocols). A RAB, as defined in TR 21.905, is the service provided by the Access Stratum to the Non-Access Stratum for transferring user data between the UE and the core network. When a RAB supports multiple sub-flows, each sub-flow carries data with specific QoS attributes — for example, one sub-flow might carry conversational speech requiring high reliability and low delay, while another carries signaling or background data with different requirements.

The sub-flow concept is particularly relevant in the context of the Iu interface user plane protocol, where the frame protocol supports multiple sub-flows within a single bearer. Each sub-flow is identified by a sub-flow number (0 to N), and the frame protocol includes sub-flow-specific fields for sequence numbers, error detection, and rate control. This allows the receiving entity to process each sub-flow independently according to its QoS requirements.

In practical terms, RAB sub-flows are most commonly associated with AMR (Adaptive Multi-Rate) speech codecs, where the speech codec produces multiple bit streams — one for each AMR mode — with different error sensitivity. The most sensitive bits (Class A) are carried in one sub-flow with strong error protection, while less sensitive bits (Classes B and C) are carried in other sub-flows with progressively less protection. This unequal error protection (UEP) approach optimizes the use of radio resources while maintaining speech quality.

The sub-flow mechanism allows the UTRAN to apply different transport channel configurations to each sub-flow. For example, sub-flow 0 might use a transport channel with a low coding rate and CRC protection, while sub-flow 1 uses a higher coding rate without CRC. This flexibility enables the network to tailor error protection to the specific needs of each data stream.

While RAB sub-flows are primarily a UTRAN concept, the underlying principle of differentiating QoS within a bearer has influenced later architectures. In 5G NR, the QoS flow concept within a PDU session serves a similar purpose, though the implementation differs significantly. The RAB sub-flow concept remains an important reference for understanding the evolution of QoS differentiation in cellular networks.

Purpose & Motivation

RAB sub-flows were created to address a specific inefficiency in early UMTS networks: the need to carry multiple service data streams with different QoS requirements over a single radio access bearer without establishing separate RABs for each stream. Before this concept, each distinct QoS requirement would necessitate a separate RAB, consuming additional radio resources and signaling overhead.

The primary motivation was the introduction of AMR speech codecs in Release 4. AMR codecs produce multiple bit streams with different error sensitivities, and the most efficient way to transmit these is through unequal error protection. RAB sub-flows provided the mechanism to apply different transport formats and error protection levels to different portions of the codec output within a single RAB.

This approach solved the problem of how to efficiently multiplex multi-rate codec streams while maintaining the appropriate level of error protection for each. It also provided a general framework for carrying any set of service data streams with differing QoS characteristics within a single bearer, making it applicable beyond speech codecs to any service requiring intra-bearer QoS differentiation.

The concept was particularly important during the early UMTS deployment phase when radio resources were scarce and efficient multiplexing was critical. It allowed operators to maximize the number of simultaneous users while maintaining acceptable quality of service for real-time applications.