ML/MC

Multilink-Multiclass PPP

Protocol
Introduced in Rel-8
A Point-to-Point Protocol (PPP) extension that enables the bundling of multiple physical links into a single logical link and supports multiple traffic classes. It enhances data throughput and provides differentiated QoS for various traffic types over Iub and Iur interfaces in UTRAN.

Description

ML/MC, or Multilink-Multiclass PPP, is a protocol defined in 3GPP for the Iub and Iur interfaces within the UTRAN architecture. It operates as an enhancement to the standard Point-to-Point Protocol (PPP), specifically designed to meet the demanding bandwidth and quality-of-service requirements of 3G and later mobile networks. The protocol functions by aggregating multiple physical transmission links, such as E1 or T1 lines, into a single logical data pipe. This aggregation is managed through a multilink protocol mechanism that fragments, sequences, and reassembles packets across the available links, thereby increasing the overall available bandwidth and providing load balancing. Simultaneously, the multiclass aspect introduces the capability to classify and prioritize different types of traffic (e.g., signaling, user plane data, O&M traffic) onto separate virtual connections or classes within this aggregated link. This is achieved by extending the PPP frame format with a Multiclass Protocol Data Unit (MC-PDU) header, which carries class identification information.

Architecturally, ML/MC is implemented in the Transport Network Layer (TNL) of the Iub and Iur interfaces, sitting above the physical layer and below the ATM or IP adaptation layers as defined in the relevant specifications. Key components include the Multilink PPP (ML-PPP) bundling function and the Multiclass (MC) framing and queuing mechanisms. The Node B (for Iub) and RNC (for Iub and Iur) must both support the protocol to establish and maintain the aggregated link. The protocol works by negotiating capabilities during the link establishment phase using Link Control Protocol (LCP) and a specific Multiclass Network Control Protocol (MC-NCP). Once established, user data packets from higher layers are classified, encapsulated with an MC-PDU header indicating the traffic class, and then passed to the multilink component. The multilink component is responsible for packet fragmentation (if necessary), assigning sequence numbers, and distributing the fragments across the member links of the bundle for transmission.

Its role in the network is critical for efficient backhaul utilization in UTRAN. By bonding multiple low-speed TDM circuits, operators could create a higher-bandwidth pipe cost-effectively without immediately upgrading to more expensive high-speed links. The multiclass functionality ensures that delay-sensitive control plane signaling receives higher priority over best-effort user data, which is crucial for maintaining call stability and handover performance. The protocol provides a standardized, reliable transport mechanism that supports the stringent delay, jitter, and loss requirements of UMTS services, forming the backbone for carrying Frame Protocol (FP) traffic between network elements. Its design allows for resilience; if one physical link in the bundle fails, traffic can be redistributed among the remaining links, albeit at a reduced total bandwidth.

Purpose & Motivation

ML/MC-PPP was created to address specific transport challenges in early 3G UMTS networks, particularly for the Iub interface connecting the Node B to the Radio Network Controller (RNC). Prior to 3G, backhaul connections often used single E1/T1 lines with basic PPP or ATM adaptation. However, the increased data rates and diverse service requirements (voice, video, data) of UMTS demanded more bandwidth and sophisticated Quality of Service (QoS) handling from the transport network. Deploying new, high-capacity leased lines (like STM-1) was prohibitively expensive for many cell sites. ML/MC provided an economical solution by allowing operators to combine multiple existing or new low-cost E1 lines, aggregating their capacity to meet the growing bandwidth needs.

The technology solved two core problems: bandwidth scalability and traffic differentiation. The multilink component solved the scalability issue by enabling 'incremental bandwidth' upgrades—operators could add E1 lines to the bundle as traffic grew. The multiclass component addressed the QoS issue by introducing a traffic management layer within PPP, which was previously lacking. Standard PPP treated all traffic equally, which was unsuitable for a network where RNC-Node B signaling (critical for mobility) must be delivered with higher priority and lower latency than user data packets. By defining separate classes, ML/MC ensured network control traffic could be protected, improving overall network reliability and performance for end-users. Its introduction in Release 5 standardized what might have been proprietary link aggregation and QoS solutions, ensuring multi-vendor interoperability in the UTRAN transport network.

Key Features

  • Aggregates multiple physical links (e.g., E1, T1) into a single logical pipe for increased bandwidth.
  • Supports multiple traffic classes (e.g., signaling, user data, O&M) for differentiated QoS.
  • Uses PPP Multilink Protocol (MP) for fragmentation, sequencing, and load distribution across links.
  • Defines a Multiclass Protocol Data Unit (MC-PDU) for identifying traffic class within frames.
  • Provides negotiation mechanisms via Multiclass Network Control Protocol (MC-NCP) during link setup.
  • Enhances transport resilience by redistributing traffic if a member link fails.

Evolution Across Releases

Rel-8 Initial

Introduced ML/MC as a defined term and protocol for UTRAN transport, specifying its use over Iub and Iur interfaces. The initial architecture supported aggregation of ATM or TDM-based physical links and defined the framing and class-based forwarding mechanisms for carrying Frame Protocol traffic.

Defining Specifications

SpecificationTitle
TS 25.424 3GPP TS 25.424
TS 25.426 3GPP TS 25.426