Stay one step ahead

Migrating to IP-based Mobile

Backhaul: Operator’s Perspective

Written by Ariel Shuper, Senior Director,

Head of Product Management & Strategy Celtro Ltd

February 2009

1
 Stay one step ahead

Migrating to IP-based Mobile Backhaul: Operator’s Perspective

Why migrate to IP/MPLS?

The evolution of mobile radio technologies has dramatically increased the bandwidth capacity of end
user equipment (UE). This poses new challenges for the mobile backhaul network because of the
significant increase in traffic volumes, and consequently the required broadband services. Planning
and dimensioning the RAN has become very difficult given the new data-oriented mobile services,
characterized by high burstiness and complex statistical behavior.

In a parallel development, establishing core networks on IP/packet-based technologies, and the

widespread use of MPLS for efficient network partitioning and segmentation have resulted in the
gradual replacement of legacy PDH/SDH networks. IP/MPLS-based networks have become
dominant not only in the core portion of the network, and a drift toward “all IP” has begun in the
metro and access networks as well, resulting in the gradual replacement of legacy PDH/SDH
devices.

The combination of increased data-centric (rather than voice-oriented) mobile capacity and of
network migration from PDH/SDH to IP/MPLS has placed significant new demands on the mobile
backhaul architecture.

The challenges of IP/MPLS mobile backhaul

The increase in 3G HSPA services (IP-based traffic) and the adoption of IP/Ethernet interfaces in the
radio elements (Node B/RNC) are reinforcing trends already followed by most telecom operators
who have been deploying residential/business packet-switched networks (PSN). In this new
situation, there is great motivation among operators to maximize revenues by converging mobile
flows over a unified multi-service PSN infrastructure. Migration of this type can reduce both CAPEX
and OPEX, gradually retire expensive legacy equipment, and achieve better utilization of the
available bandwidth. To accomplish this migration, however, several technological challenges must
be solved first:
• Several types of base stations must be backhauled simultaneously using TDM transport in
the BSS and ATM transport in the UTRAN
• Remote OAM, fault isolation, performance monitoring, and measurement capabilities

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 2

 Stay one step ahead

• Carrier grade reliability and guarantee of advanced QoS

• Strict synchronization capabilities

The benefits of IP-based mobile backhaul

What is the incentive of mobile operators to migrate to IP-based backhaul? The short answer is: to
keep up with the sharp increase in traffic volume. With each successive generation of radio
technology (the omnipresent Gs), over-the-air rates have increased dramatically, and with them the
range of available end-user applications. HSPA (3G), and soon HSPA+ and LTE (4G) are opening
the door to mobile broadband services that generate unprecedented volumes of data, exceeding the
capabilities of the traditional networks that were designed to carry voice data.

According to Heavy Reading's “3G network evolution to LTE,” by mid-2007, larger operators offering
3G HSDPA service in the five big European markets were transporting 1,000 GB per day as a
baseline, with substantially higher peaks. In September 2007, Ericsson released figures on the
average throughput of its Radio Network Controllers (RNCs, which typically support 200-300 Node
Bs). Ericsson's figures, shown in the chart below, are considered representative of the worldwide
average.

Source: Ericsson Strategy & Technology Summit (London, September 2007)

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 3

 Stay one step ahead

The chart shows that total traffic volume increased by ~50% to 37.5 Gbit/RNC/hour, packet traffic
increased by ~100% percent to 20 Gbit/RNC/hour, and HSPA traffic increased by ~240% to 12
Gbit/RNC/hour.

To achieve the low-cost, low-latency, flexible network that can carry the mobile broadband services
operators are eager to offer, they need to adopt an IP-based mobile backhaul network
architecture. The benefits of such a migration are many.

• Simplified networks using a small number of elements to save both OPEX and CAPEX

• The ability to move high volumes of data without a commensurate increase in the cost of
service delivery

• A flexible network that can support both mobile and generic IP access networks

• Mobile broadband services that are competitive with wired broadband services in
price/performance

Typical migration scenarios

Given the broad mix of technologies and architectures used by mobile carriers, there are many
possible migration scenarios. We illustrate here three typical cases of migration to IP:

• MPLS over new and legacy transport

• GSM (Abis/Ater) over IP

• Carrier Ethernet

Celtro solutions, which support all technologies and architectures, allow operators to choose the
migration path that best suits them, based on their legacy equipment and technologies.

Migration to MPLS over new and legacy transport

Migrating Radio Access Network (RAN) to packet-switched technologies is one way in which mobile
operators can increase network flexibility and reduce operating costs. Operators generally prefer to
do it gradually, to avoid having to “forklift” or “truck roll” their existing backhaul infrastructure. At the
same time they must protect the quality of their voice services, which still generate higher revenues
than the data services despite the reverse ratio in traffic volumes.

MPLS is at present the preferred technology used to meet these goals. MPLS is a packet-switched
technology that enables efficient backhaul of data-based services using existing transport resources

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 4

 Stay one step ahead

and providing secure and protected methods to ensure that voice services are maintained at the
same level of user experience.

Migrating to an MPLS-based backhaul network requires defining dedicated virtual tunnels to

connect the radio network elements (Node B/BTS with RNC/BSC-MSC). The tunnels, called Label
Switched Paths (LSP), contain additional virtual tunnels (called Pseudo Wires) which are
differentiated by QoS and used to backhaul the mobile traffic (voice, data, and signaling) generated
at the NobeB/BTS and RNC/BSC. Pseudo wires (PWE) are used to carry mobile services through
the designated LSPs. The function of the PWE is to emulate legacy traffic such as ATM (RFC 4717)
and/or TDM (RFC 4553/5086), which the radio network elements produce in a packet-based format.
After the various types of traffic are emulated, they are tunneled into the appropriate paths. An
additional benefit of the tunneling method is the creation of a protection scheme to assure service
continuity. The typical protection scheme is based on the RSVP-TE protocol (RFC 3209), which
creates a standby LSP in parallel to the working LSP to ensure service continuity (within a 200ms
switching time) in case of failure of one or more network elements along the path between the
NodeB/BTS and the RNC/BSC.

A significant advantage of MPLS in mobile backhaul networks is their ability to reuse existing
network elements that are not packet-switch oriented. Typical mobile backhaul networks contain a
significant number of legacy PDH/SDH network elements installed in the past for ATM/TDM
transport. MPLS can be used over native Ethernet elements alongside PDH/SDH legacy elements.
MPLS requires an intermediate protocol such as PPP or Multi-Link PPP (RFC 1990) when using
PDH links and Packet over SDH (RFC 1619) when using SDH links. At the cost of a relatively small
overhead, these protocols allow to transport MPLS-based traffic over legacy network elements,
which in return creates significant saving through the reutilization of the transport infrastructure.

The figure below illustrates a solution based on Celtro technology for migrating 2G and 3G traffic to
an MPLS-based network. TDM/ATM traffic is emulated at the cell-site by Celtro’s DMT 1000, which
carries it over Ethernet or PDH/SDH links using various MPLS PW/LSPs with differentiated QoS,
according to the Differentiated Services mapping model. At the edge of the backhaul network,
Celtro’s DMT 4000 receives the MPLS stream and terminates the MPLS session, de-encapsulating
the various traffic types into their original format (2G-TDM, 3G-ATM) and forwarding them to their
destination (2G-MSC, 3G-RNC):

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 5

 Stay one step ahead

MPLS migration in a mixed backhaul network using Celtro technology

This scenario has been successfully implemented with Celtro equipment by large operators in
Europe and Africa.

GSM (Abis/Ater) over IP

Some operators have been migrating their backhaul to IP not only to accommodate high 3G data
rates but also to lower backhaul costs. This is the motivation of mobile operators who migrate their
GSM backhaul to packet-switched solutions. With the rapid introduction of Ethernet lease lines and
packet-switched infrastructure, the transport cost of packet-switched traffic have become more
attractive, especially when the transport provider is a sister or parent company of the mobile
operator.

Several technical issues must be addressed to use packet-switched backhaul for GSM services.

GSM traffic is based on TDM technology, with a deterministic bandwidth consumption that is not
proportional to the actual bandwidth being used. Adapting the bandwidth consumption to the actual
required bandwidth reduces dramatically the required capacity, and with it the backhaul costs. To be
transported as packet-switched traffic, GSM traffic requires pseudo wire (PWE) emulation. IETF
defines two different PWE options suitable for GSM services: SAToP (RFC 4553) and CESoPSN
(RFC 5086). These options emulate TDM traffic (without the recommended adaptation of the actual
bandwidth to the consumed bandwidth) as packet-based traffic in various formats. SAToP is

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 6

 Stay one step ahead

designed for packet-based conversion of full E1s, CESoPSN performs packet-based conversion of
Nx64kbps (specific time-slots).

In addition to traffic emulation as packet-based traffic (with or without bandwidth adaptation), it is

necessary to define an MPLS-based LSP as a point-to-point logical connection between each BTS
and the aggregating BSC, and between each BSC and the responsible MSC. These LSPs ensure
that all packets are transported along a direct path, without the disorder, packet loss, or
differentiated delay that can occur in packet-based networks. As explained above, the logical path
architecture allows adding the protection mechanism that ensures the service availability and
network reliability necessary for voice services.

Another significant challenge in migrating to a packet-based architecture is meeting the mobile

requirement for accurate network synchronization of the radio elements (BTS/BSC). Typically, the
radio elements obtain the required frequency synchronization from the E1/STM-1 (PDH/SDH)
interfaces (G.823). When migrating to packet-switched technologies, which are inherently
asynchronic, there is a need to provide adequate synchronization for all elements in the network.

Several methods have been suggested for expanding packet-based technologies (Ethernet,
IP/MPLS) to distribute frequency and clock information. IEEE developed a protocol for precision
clock synchronization. The IEEE-1588v2 enables sub-microsecond synchronization of clocks by
using a master clock that sends multicast synchronization message frames containing timestamps.
All IEEE-1588v2 receivers correct their local time based on the received timestamp and an
estimation of the one-way delay from transmitter to receiver.

Another timing method, called Network Time Protocol (NTP), was defined by IETF to be used to
distribute clock time over IP networks. Time servers send timestamps in packets, usually in
response to requests from a client that may be in contact with several different time servers. New
ITU-T recommendations attempt to define additional methods to transport timing reference over
Ethernet networks. The ITU-T G.8261 standard defines a new timing method called Synchronous
Ethernet, designed for commonly used dedicated-media full-duplex Ethernet interfaces, in both
copper and optical physical layers that transmit continuously. At the physical layer, Ethernet
(similarly to SDH) adopts the serial code stream in transmission. Data flows are transmitted from the
Ethernet interface with a high-accuracy clock, and the receiver can recover and extract the clock and
maintain the high-accuracy performance at the same time.

All the above mechanisms, including traffic emulation, bandwidth adaptation, MPLS LSP settings,
and network synchronization are essential for enabling GSM (Abis/Ater) backhaul over packet-

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 7

 Stay one step ahead

switched technologies. Celtro has developed innovative solutions for coping with all the challenges
listed above. The figure below illustrates a GSM over IP solution using Celtro's all-in-one-box
switching solution, with the new DMT 1000 (DPS family) at the base stations and DMT 4000 at
BSCs.

GSM (Abis/Ater) backhaul over IP/MPLS transport using Celtro technology

This solution has been successfully implemented with Celtro equipment at a large African and FSU
Operators.

Carrier Ethernet in mobile backhaul networks

Mobile operators often want to share the resources of a parent/sister company that has wire-line
network assets. This can produce significant savings because it eliminates the need to build and
maintain two separate networks in parallel. Moreover, recently installed radio elements
(NodeB/RNC) have native Ethernet switching capabilities that allow them to merge with the wire-line
networks that have already migrated to Ethernet-based networks long ago.

In the converged network, each service (wire-line and wireless) uses a dedicated Virtual Private
Network (VPN), a method that allows traffic separation to avoid congestion and breaches in network
security that can occur between different networks/services. There is ongoing debate about the

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 8

 Stay one step ahead

preferred method of network virtualization, and whether it should be based on layer 3 (L3VPN) or
layer 2 (L2VPN) mechanisms. When merging mobile and wire-line networks, Layer 2 VPN appears
to be more suitable for the mobile backhaul network architecture, although there are exceptional
cases in which L3VPN is preferable. .

Among the different L2VPN options, Virtual Private LAN Services (VPLS), or more precisely the
Hierarchical-VPLS (H-VPLS) architecture seems to meet the needs of the current mobile backhaul
network. VPLS allows mobile operators to create a virtual LAN environment between their radio
elements (Node B/BTS and RNC/BSC). H-VPLS partitions the network into several domains that are
interconnected by means of an MPLS core. Considerations for edge devices at the cell-site are thus
simplified, requiring either simplified MPLS/IP capabilities (route to the local n-PE) or Ethernet
switching mechanisms such as VLAN Tag Stacking (Q-in-Q, IEEE 802.1ad provider bridges). Using
Ethernet as the edge technology simplifies the operation of the edge domain and reduces the cost of
the edge devices dramatically.

When using Ethernet mechanism such as VLAN Stacking, each Node B receives a dedicated VLAN
tag that specifies traffic origination or destination. The various services/traffic types generated at the
cell-site receive their respective VLAN tags, which are shared between similar traffic types/services
throughout the backhaul network. Having a dedicated VLAN for every service throughout the
network makes it possible to create service differentiation within the network, while using the 2nd
VLAN tag to identify the source/destination of each service for more efficient transport and reduced
delay. MPLS is still required for H-VPLS because it creates the virtual tunnels (LSPs) throughout the
network and enables service differentiation. Advancements in Ethernet technology (e.g., PBT,
PBBT-TE) further improve this model (for example, eliminating the need for MAC learning in the
network, expanding VLAN tag options, etc.).

The figure below illustrates how using the Celtro DMT family of products allows a VPLS network to
support mobile backhaul services, sharing the Metro Ethernet infrastructure with wireless networks.
In the illustration, the different VLAN networks represent different services or traffic types that are
carried throughout the Metro Ethernet network between the Node B/BTS and RNC/BSC with
underline MPLS tunnels.

This solution has been successfully implemented by a large operator in Russia using Celtro
equipment.

Migrating to IP-based Mobile Backhaul: Operator’s Perspective 9

 Stay one step ahead

Carrier Ethernet backhaul in converged networks using Celtro technology

Summary

The higher demand for service delivery and bandwidth produced by 3G applications places new
challenges before mobile operators To be able to offer the profitable broadband data services that
customers demand and that the RAN is already making possible, mobile operators must address
several issues on the backhaul side. For adequate service delivery they must be able to ensure
broadband with sufficient capacity to handle peak traffic, including future growth. They must also
acquire the flexibility to handle both legacy and current technologies, ATM alongside IP/MPLS. And
they must be able to add new services without the added cost of new infrastructure. Migration to a
packet-switched network (PSN) is the key to meeting these challenges. But given the many
infrastructures and technologies that operators have adopted over the years, migrating to PSN while
protecting their investment in the legacy equipment is a challenging task.

Migrating to IP-based Mobile Backhaul: Operator’s Perspective

10
 Stay one step ahead

© Copyright by CELTRO LTD. 2009. All rights reserved worldwide.

The information contained in this document is proprietary and is subject to all relevant copyright, patent and other laws
protecting intellectual property, as well as any specific agreement protecting the rights of CELTRO LTD. in the aforesaid
information. Neither this document nor the information contained herein may be published, reproduced or disclosed to
third parties, in whole or in part, without the express, prior, written permission of CELTRO LTD. In addition, any use of
this document or the information contained herein for any purposes other than those for which it was disclosed is strictly
forbidden.
CELTRO LTD. reserves the right, without prior notice or liability, to make changes in equipment design or specifications.
Information supplied by CELTRO LTD. is believed to be accurate and reliable. However, CELTRO LTD. assumes no
responsibility for the use thereof or for the rights of third parties that may be affected in any way by the use thereof.
Any representation(s) in this document concerning performance of CELTRO LTD. product(s) are for informational
purposes only and are not warranties of future performance either express or implied. The CELTRO LTD. standard
limited warranty, stated in its sales contract or order confirmation form, is the only warranty offered by CELTRO LTD. in
relation thereto.
This document may contain flaws, omissions or typesetting errors; no warranty is granted nor liability assumed in relation thereto unless
specifically undertaken in the CELTRO LTD. sales contract or order confirmation. Information contained herein is periodically updated
and changes will be incorporated into subsequent editions. If you have encountered an error, please notify CELTRO LTD. All
specifications are subject to change without prior notice.

Migrating to IP-based Mobile Backhaul: Operator’s Perspective

11