BROADBAND: Road To Efficiency

Comment

Digg

Del.icio.us

Today, many mobile operators face the challenge of upgrading their infrastructure to support growth in new mobile services and, at the same time, cutting costs.

The response: a new focus on eliminating inefficiencies in their network designs. One such area of inefficiency is the legacy layer 2 switching and transport infrastructure which supports ATM and FR connectivity, required as part of the GSM/3G-3GPP standards.

MPLS-based designs are an attractive alternative to this legacy infrastructure. We will examine the migration to MPLS, and see what are its short- and long-term advantages from the perspective of GSM-GPRS operators as well as that of new 3G WCDMA requirements.

GPRS: Today’s Transport
For many mobile GPRS operators, one area of inefficiency is in the transport of GPRS data traffic from the many base station controller (BSC) sites back to the centralized serving gateway support nodes (SGSN).

Controlling costs in this part of the network is increasingly challenging as more services are activated, and also as operators prepare for EDGE. EDGE will also increase the bandwidth requirements, with data rates of up to 384 kbps. Additionally, new GPRS-based services such as MMS and introduction of push-to-talk will increase bandwidth requirements in the access network.

In the 3GPP standard design for GPRS, the Gb interface between the packet control units (PCU) and SGSNs uses frame relay. This is typically implemented using multiple E1 interfaces on both BSC/PCU and SGSN, as per the diagram below.

In many networks, there are multiple BSC/PCU sites, which aggregate into a small number of central sites where the GPRS support nodes are located. In these networks, inter-exchange connectivity uses PDH and SDH transmission capacity, requiring numerous E1 channels to be nailed up. This approach is well understood but it is also inefficient.

The key problems are the lack of statistical multiplexing on the data traffic and high internal cost of using numerous numbers of E1 circuits delivered through traditional transmission equipment. In addition, each incremental upgrade or new site may require ordering of additional circuits, leading to increased operational overheads and longer planning cycles.

Invest in Frame Relay/ATM?
One option that mobile operators may consider is to concentrate Gb traffic using a traditional frame relay/ATM network. Whilst providing statistical multiplexing, this approach increasingly has a number of drawbacks.

n Limited flexibility in inter exchange and inter-site connectivity; restricted to traditional PDH/SDH WAN interfaces only (for example, no Ethernet)

n Fixed ATM cell overhead can be inefficient in the case where FR circuits are inter-worked to ATM

n No native IP/MPLS capabilities; this reduces investment protection and also means other IP traffic used by the mobile operator between those sites must use a separate network infrastructure (e.g., OSS- or billing-traffic based on IP)

n Traditional FR/ATM switches now have a limited useful lifespan, as the amount of research and development into these platforms continues to taper off

MPLS Optimizes Evolution
A more effective alternative is to migrate to a modern MPLS solution to transport layer 2 FR circuits.

MPLS has successfully proven itself to not only handle IP services such as routing and L3 VPNs but also layer 2 transport, for a variety of services including frame relay, ATM, and Ethernet. Layer 2 transport over MPLS works by the efficient encapsulation of layer 2 frames within MPLS at the provider edge (PE) nodes. The core technology itself has been standardized for some time, with multi-vendor interoperability a reality today.

An Optimized Evolution Using MPLS

The MPLS solution takes advantage of modern routing platforms, capable of handling a variety of both layer 3 and layer 2 services natively, with no performance degradation or additional hardware investment required. Additionally these platforms have rich QoS mechanisms to ensure that layer 2 service characteristics are maintained.

The advantages of an MPLS solution are numerous.

n Cost Savings: By consolidating layer 2 and layer 3 services, savings are made by significantly reducing duplicate investment in both TDM/FR/ATM switches and IP routers.

n The uplink from BSC sites can use any interface type supporting MPLS; including traditional WAN interfaces such as DS-3, STM-1, and STM-4 (both ATM cell and PoS packet variants) as well as Ethernet interfaces. For example, long haul GE over dark-fiber could be used to natively link sites up to 80 km apart—providing increased bandwidth at a substantially lower cost.

n Operational Savings: By running one multiservice infrastructure, operations can be streamlined, and also day-to-day dependencies on other groups for additional transmission capacity can be reduced.

n MPLS is likely to be already deployed as a key core data network technology, and is already well understood by the carriers.

n An MPLS solution built on a modern routing platform is in alignment with technology evolution to 3G, with IP centric designs standardized by the 3GPP. This provides investment protection and future proofing.

Investment Protection
In the initial 3G WCDMA phases with 3GPP Release 99 and Release 4, the Iu ps interface between the RNC and the SGSN for data services between the 3G radio network controller (RNC) and the SGSN uses ATM AAL5.

The IP/MPLS network platforms supporting GPRS FR transport between BSC/PCU and SGSN can be used to transport 3G Iu ps user and signaling traffic, using identical layer 2 MPLS network technology, saving costs.

In addition to the data interfaces, voice services are also implemented over ATM using AAL2 traffic class. ATM over MPLS technology can also be used to transport voice services.

Far from being just a theoretical design, a number of new 3G deployments are indeed using this technique with an MPLS infrastructure, rather than having to build out new or expand existing ATM networks to support new 3G voice, video, and data services.

3G Evolution to Pure IP
The initial 3G deployments based on 3GPP Release 99 and Release 4 standards will use ATM connectivity either delivered on ATM or MPLS infrastructure. From Release 5 onwards mobile operators may move to native IP protocols in their mobile networks.

One of the key enhancements in Release 5 is the introduction of the IP transport option in the UTRAN. Whilst in Release 99 and 4, ATM connections are used at the transport layer, Release 5 offers the possibility of using native IP as the interface in the Iub, Iur, IU-Ps, and Iu-Cs interfaces, i.e., all the traffic: data, voice, and signaling.

This enables use of a simplified network design built around IP and using common routing platforms, which can simultaneously handle layer 2 services with MPLS. The IP UTRAN solution also enables flexible use of different link technologies, ranging from traditional TDM interfaces such as E1 and nx64k through to packet oriented interfaces such as FE/GE if Metro Ethernet access is available between sites. This is relevant as the other big enhancement in Release 5 is the introduction of the IP Multimedia Subsystem, allowing native VoIP/IP multimedia support directly from the client terminal.

Today, mobile operators can lower ongoing costs and better accommodate future requirements by consolidating layer 2 traffic over an MPLS-based solution. This also prepares mobile operators effectively for a future evolution to native IP in the mobile network.

Layer 2 Requirements in the 3G Network

This approach mirrors the trend at the core of many large wireline operators looking to cap investment in core ATM switches and consolidate on a converged multiservice MPLS backbone.

Far from being a ‘demonstration’ technology, layer 2 MPLS services have been successfully deployed in large production networks. Korea Telecom, Verizon, and MCI are benefiting from deploying layer 2 services on Juniper Networks’ routing platforms—proving their maturity and showing the way forward for operators building out their networks for the future.

Simon Newstead, product manager-mobile, voice, and EDGE, Juniper Networks (Asia Pacific)