Operators are investing now in multi-service, high bandwidth backhaul solutions. How can they ensure operational flexibility, meeting current and future demands?
If you talk to those on the supply side, there's little doubt that operators are now really beginning to see increased 3G usage drive high bandwidth services. This in turn has meant that the old ways of carrying traffic from base stations back to the RNC and into the core network just won't scale. A typical 2G base station serving an average population of 1000 subscribers requires a connection of only a few megabits (or a few E1/T1 lines). But with HSPA driving usage, and 4G around the corner, operators are looking at massive growth in backhaul capacity requirements.
At the moment, there is agreement that operators will be spending money now in order to avoid spending more later. That means that operators know they are moving to an all-IP network in the core and RAN, using Metro Ethernet in some way to provide the transport. But there is hot debate about the best way to arrive at the goal, and whether all the barriers to that vision have been overcome.
Infonetics Research says that mobile operators and backhaul transport providers spent $3.7 billion worldwide on mobile backhaul equipment in 2007, and are expected to increase their spending in the high double-digit percents from 2009 to at least 2011. This is encouraging the vendors of Ethernet based network equipment.
"All market indicators support continued growth of the mobile backhaul market. Manufacturers and service providers have had residential broadband and corporate services as the main thrust of their businesses for a long time, and now mobile backhaul makes up a third area that nearly all of them are focusing on. We expect to see the Ethernet mobile backhaul revolution really kick off in 2009," said Michael Howard, principal analyst of Infonetics Research and lead analyst on the report.
Infonetics highlights three major factors forcing a migration to packet backhaul: the first is the increasing numbers of mobile subscribers, reaching 4.4 billion worldwide in 2011. The second, as we have mentioned, is an explosion in mobile data and video use, requiring providers to significantly increase bandwidth offerings rapidly. The third is the growth of competition, forcing operators to upgrade their network capacity to improve and add new subscriber services. These upgrades will include IP/Ethernet BTS/NodeBs, WiMAX, and LTE New cell site backhaul connections, which drive equipment spending that will roughly quadruple worldwide from 2007 to 2011.
Crucial to this is the forecast that the IP/Ethernet portion of worldwide mobile backhaul equipment revenue is set to skyrocket, racking up a triple-digit five-year compound annual growth rate from 2007 to 2011.
By 2011, service providers using PDH, ATM over PDH, or SONET/SDH for their mobile backhaul connections will be paying roughly 3 to 40 times as much in service charges per connection as those using Ethernet, DSL, coax cable, or PON, Infonetics estimates.
There is no single technology solution that can alleviate the backhaul bottleneck as mobile networks migrate to 3G and HSPA technology and services. Rather, a mixture of technologies and solutions are required to address the challenge. Technologies addressing the backhaul bottleneck may be divided into two categories:
1. Technologies that increase backhaul infrastructure capacity
2. Solutions that enhance network performance through traffic aggregation and optimization – maximizing traffic and service delivery over existing infrastructure
Since there is no single solution for the backhaul bottleneck, operators will be using a mix of solutions from each of these two categories.
Increasing Capacity through Infrastructure
When considering increasing backhaul infrastructure capacity, there are different solutions for the leased backhaul network and the owned or self-built network.
For the backhaul network based on leased capacity, adding conventional TDM leased lines (E1/T1 and STM1/OC3) may partially address the backhaul challenge at the cost of a significant increase in network OPEX and delay in the introduction of new services due to lead time of additional leased E1/T1's.
New wireline access technologies, like xDSL and Metro Ethernet, may also be used as a cost effective alternative to backhaul cellular traffic. Initially, these services were planned to carry only best effort mobile data services, but they will migrate in the future to provide a converged IP RAN for all mobile services.
For the owned or self-built network, several solutions are available to increase backhaul capacity. Evolution of PTP microwave with adaptive modulation, new PTMP microwave technologies and WiMax are some of the alternatives to increase capacity on microwave backhaul infrastructure. The ultimate solution is, of course, deployment of fiber down the access network close to the cell sites. However, this solution is often unviable, since today's fiber technologies can provide bandwidth relief only with huge up front investments and very long lead times to deployment.
So until they can reach this networking nirvana, most operators are looking at a hybrid approach to mobile backhaul, keeping 2G and 3G voice on current TDM technology, and using packet technology for the growing data service EV-DO, EDGE, and HSDPA traffic. The T-Mobile, Swisscom Mobile, and Telecom Italia contracts for IP, Ethernet, and pseudowire cell site backhaul are the first of many to come over the next 18 months.
Steve Dyck, Director of Mobile solutions at Alcatel, says that operators can choose to upgrade where necessary, whilst keeping future demands in mind.
"Where they can, they can go for Ethernet backhaul. And even if it's not something that they can get to every cell for a long time, they can get to the hub sites, using TDM products for immediate growth at the individual BTS. Then with the IP based platforms being deployed in 2008. As RAN equipment moves to IP, then operators have the equipment they need in pace already to support that move.
"Operators that adopt a hybrid approach can really take three main approaches. The first is BTS that have an Ethernet port as well as an ATM port. HSPA traffic can be carried natively over Ethernet and transported to the network. With TDM traffic bring carried over TDM. The second option is where you have a BTS with no Ethernet port, possibly where legacy equipment has not yet been upgraded. In this case you need a device at the cell site that can peel the different traffic off and then send it over Ethernet. Another option is to carry all traffic directly onto a Carrier Ethernet network, with legacy interfaces supported with an interworking function or circuit emulation service."
The demands of the latter approach, is that it requires stringent timing and synchronisation, high availability, scalable bandwidth and hard QoS. Yet because of the problems it solves, and the potential of greatly reduced operational cost, that latter option is the one that nearly all operators are interested in getting to.
Reduced OpEx resulting from operating a single network rather than several parallel networks would be derived from improved statistical gain from statistically multiplexing all backhaul traffic onto one network; Maximum network flexibility to handle real time traffic demand changes, as well as long term traffic mixture changes, by dynamically sharing converged network resources across all mobile services.
In the past, TDM switches and access cross-connects were used in 2G networks. Today, with the rollout of 3G technologies, packet switches based on ATM or Ethernet/IP technologies are used. Some of these switches use traffic optimization in addition to standard layer 2 aggregation, in order to multiply the network capacity delivery performance. Such switches can also provide dynamic bandwidth allocation for all services, increasing network flexibility by handling real time variation in traffic demand and long term transition from traditional 2G to new 3G and HSPA services. These switches support QoS in order to properly allocate network resources, maximizing service delivery and ensuring operator revenue streams.
Timing issues
Timing over Packet (ToP) began as a sideshow of the Circuit Emulation wave that flooded the telecom industry in the early 2000s. Reinforced by the emerging new PWE3 family of pseudowire (PW) standards in the IETF, more and more carriers began looking for cheaper ways to carry their legacy TDM services, giving up traditional expensive leased lines.
Back in those days, ToP was merely considered a way to enable timing "transparency" across a physical layer asynchronous network (today called Adaptive Clock Recovery or ACR). The term "transparency" is placed in quotation marks, as the performance of early ToP mechanisms fell short of what was needed to ensure true end-to-end clock transparency. This has radically changed today as modern ToP mechanisms are rapidly approaching the performance envelope of their SDH/SONET counterparts, enabling conformance to the latter's most stringent timing requirements.
"As the experts in Carrier Ethernet demarcation, we offer a comprehensive portfolio that allows carriers and providers to start a service over fiber, terminate it over bonded SHDSL.bis and manage it end-to-end, even in cases where third party networks are involved, without affecting the end user's experience,"states Ron Agam, Director of Product Management at RAD Data Communications. "RAD's EtherAccess technology provides carriers and service providers with the ability to deploy lower cost, highly reliable standardized Ethernet services according to predefined SLAs regardless of the underlying access infrastructure or transport network."
"The industry clearly recognizes that ATM pseudowires over packet networks are a key requirement in supporting 3G and 3.5G traffic such as HSPA, and this is another area in which RAD's ACE cell site gateways were tested successfully," Agam continues. "Cell site gateways are proving to be a necessity, especially for transport providers and operators with self-built networks," he concludes. "Depending on the capabilities of specific NodeBs, intelligent cell site gateways provide the only technologically viable solution."
"Mobile operators are facing a significant spike in bandwidth demands in the backhaul due to the proliferation of 3G and eventually 4G-based data services and the emergence of high-speed air interface enhancements such as HSPA," said
Michael Howard, principal analyst at Infonetics Research. "At the same time, these operators are seeking to dramatically reduce their operating costs and are looking at migration from separate, legacy ATM and TDM networks to more cost-effective converged IP/MPLS-enabled multi-purpose networks. During this migration, mobile backhaul networks have become the bottlenecks in these new network architectures, and the work from the IP/MPLS and Metro Ethernet Forums offers viable solutions to the problem."
The goal of the IP/MPLS Forum's work on Mobile Backhaul is to provide guidelines on the architecture and technology choices for IP/MPLS RAN backhaul within the various network environments (legacy, IP, converged). The IP/MPLS Forum's MPLS Mobile Backhaul Initiative (MMBI) proposes a framework for the use of MPLS and Ethernet technology to transport RAN backhaul traffic over access, aggregation and core networks. The framework includes a reference architecture that is intended to be broad enough to cover all possible deployment scenarios, providing recommendations on how to deploy MPLS in each of these scenarios. This will create a reference guide that will allow vendors and operators to select the appropriate feature sets for their specific scenario.
The focus is on a shared network infrastructure that is able to support (or replicate) existing legacy services (2G, 2.5G) as well as new services based on 3G and 3.5G (Eg: HSPA). Emerging technologies such as LTE, mobile WiMAX and UMB and are also considered. This will enable a migration path between existing legacy ATM and TDM backhaul networks to a more cost-effective, converged, MPLS-enabled, and multi-purpose network.