Asynchronous Transfer Mode

Description

Asynchronous Transfer Mode (ATM) is a high-speed, connection-oriented switching and multiplexing protocol standardized by the ITU-T and adopted by 3GPP for the transport infrastructure of 2.5G and 3G networks, particularly in the UMTS era. Unlike packet-switched networks using variable-length frames, ATM segments all data into small, fixed-size 53-byte units called cells. Each cell consists of a 5-byte header and a 48-byte payload. The header contains critical control information, including Virtual Path Identifier (VPI) and Virtual Channel Identifier (VCI) fields used for routing, a Payload Type Identifier (PTI), Cell Loss Priority (CLP) bit, and Header Error Control (HEC). This fixed cell structure enables predictable latency and efficient hardware-based switching, which is crucial for real-time services like voice and video.

In the 3GPP architecture, ATM served as the primary Layer 2 transport technology for both the Core Network (CN) and the Radio Access Network (RAN). Within the UTRAN (UMTS Terrestrial Radio Access Network), the Iub interface between the Node B and Radio Network Controller (RNC), the Iur interface between RNCs, and the Iu interface between the RNC and the Core Network all utilized ATM as the underlying transport. User plane data and control plane signaling (e.g., RANAP, NBAP) were carried over ATM Adaptation Layer (AAL) protocols. AAL2 was specifically optimized for delay-sensitive, variable-bit-rate traffic like compressed voice in circuit-switched calls, while AAL5 was used for bursty data traffic and signaling messages.

The protocol operates by establishing virtual circuits (VCs) between endpoints before data transfer begins. These can be Permanent Virtual Circuits (PVCs), which are statically configured, or Switched Virtual Circuits (SVCs), which are dynamically set up and torn down via signaling. The connection-oriented nature, combined with Traffic Management and QoS mechanisms defined by the ATM Forum (e.g., Constant Bit Rate (CBR), Variable Bit Rate (VBR), Available Bit Rate (ABR)), allowed network operators to guarantee specific bandwidth, delay, jitter, and loss characteristics. This made ATM exceptionally suitable for the multiservice requirements of early 3G networks, which needed to simultaneously support traditional circuit-switched voice and emerging packet-switched data services with strict QoS demands.

ATM's role extended to the Core Network, where it was used in the circuit-switched (CS) domain for transporting voice traffic and in the early GPRS packet-switched (PS) domain as a transport option for the Gn and Gp interfaces between GSNs (GPRS Support Nodes). Its integration into 3GPP was comprehensive, covering user plane transport, control plane signaling transport, and network management. However, its complexity, relatively high overhead due to the cell header (about 9.4%), and the industry's broader shift towards 'All-IP' networks led to its gradual phase-out in favor of Ethernet and IP/MPLS in later 3GPP releases.

Purpose & Motivation

ATM was developed to create a unified, high-performance networking technology capable of integrating diverse traffic types—voice, video, and data—on a single infrastructure. Prior to ATM, telecom networks used separate systems: circuit-switching for voice (with guaranteed quality but inefficient for bursty data) and packet-switching like X.25 for data (which was flexible but too slow and unreliable for real-time voice). The goal of ATM was to combine the best of both worlds: the deterministic, low-latency performance of circuit-switching with the efficiency and flexibility of packet-switching. This integrated services digital network (ISDN) vision was critical for the emerging broadband ISDN (B-ISDN) and aligned perfectly with 3GPP's need for a robust transport backbone for UMTS that could handle the mixed traffic profiles of mobile services.

When 3GPP standardized UMTS in Release 99, a key requirement was a transport network that could provide stringent QoS for real-time conversational voice and video calls while efficiently handling interactive and background data sessions. ATM, with its mature standards, proven hardware, and sophisticated traffic management capabilities, was the natural choice. It solved the problem of how to build a scalable RAN backhaul that could meet the delay and jitter specifications for compressed voice (AMR codec) over a packetized infrastructure. Its connection-oriented nature provided the necessary traffic engineering and admission control to prevent network congestion and ensure service level agreements.

The adoption of ATM addressed the limitations of the purely IP-based internet, which, at the time, lacked robust, standardized QoS mechanisms (IntServ/RSVP was complex and not widely deployed). ATM provided a controlled, telco-grade transport layer upon which IP services could be reliably overlaid. It enabled the early commercialization of 3G services by offering a stable and predictable transport technology that network equipment vendors and operators were already familiar with from fixed-line broadband deployments. However, its purpose was ultimately transitional, as the long-term vision always pointed towards a simplified, cost-effective, and ubiquitous IP-based transport layer.

Key Features

• Fixed-size 53-byte cell structure for predictable switching latency
• Connection-oriented virtual circuits (PVCs/SVCs) for guaranteed resource allocation
• Comprehensive QoS classes (CBR, rt-VBR, nrt-VBR, ABR, UBR) for traffic management
• Segmentation and Reassembly via ATM Adaptation Layers (AAL2 for voice, AAL5 for data)
• Integrated support for both circuit-switched and packet-switched service transport
• Hardware-based switching enabling high-speed, low-latency forwarding

Evolution Across Releases

Defining Specifications