IEEE 1588 Timing for Mobile Backhaul: The Road to Plug & Play

Transcription

1 White Paper IEEE 1588 Timing for Mobile Backhaul: The Road to Plug & Play Prepared by Patrick Donegan Senior Analyst, Heavy Reading on behalf of June 2011

2 Ethernet Backhaul Deployment Is In Full Swing Mobile operators throughout the world are now turning up Ethernet backhaul to live commercial service in large volume. After many years of trials, the leading operators are mastering Ethernet backhaul. It now works well not just for data services, but increasingly for voice services, too and this has been a factor contributing to the brighter revenue and margin outlooks that leading operators in some regions have been reporting, even in a difficult global economic environment. The challenge now is for these leading operators to push Ethernet backhaul out beyond the initial deployment phase, and for other operators around the world to follow their lead in rolling it out. Heavy Reading predicts that the proportion of the world's cell sites in live service with Ethernet backhaul will increase from 6 percent at the end of 2010 to 75 percent by the end of Given the well-understood potential of the backhaul to serve as a bottleneck with respect to mobile broadband subscriber growth or profitability or both achieving this trajectory, or something close to it, is pretty much mandatory for mobile operators. While the introduction of Ethernet backhaul has presented a major technological and operational challenge for mobile operators, most of them have nevertheless fallen back on legacy TDM capabilities in the backhaul for network timing or synchronization. DS1s or E1s continue to be easily the most popular approach to backhaul synchronization, even where Ethernet service is deployed. According to the March 2011 edition of Heavy Reading's Ethernet Backhaul Quarterly Market Tracker, 43 percent of Ethernet backhaul deployments were still being supported by DS1 or E1 synchronization at the end of In the remaining 57 percent of deployments, backhaul synchronization was still fragmented across a variety of different industry standards and proprietary vendor solutions. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 2

3 While operators might continue to rely on TDM timing for Ethernet backhaul initially, as mobile data traffic grows and Ethernet backhaul matures, leaving TDM at the site becomes an unnecessary cost TDM Timing Is Not a Long-Term Solution The reason that TDM timing remains the operator's preferred solution even in an Ethernet backhaul environment is that it is mature, familiar and readily available. In the case of most operators that have deployed a hybrid backhaul architecture in which voice continues to be carried over TDM but high-speed data is transported over Ethernet using some E1s for timing has been, and still is, a no-brainer. Even for operators that have chosen a pure Ethernet backhaul solution for both voice and data, many continue to leave a single E1 or DS1 at the cell site exclusively to deliver timing again because it's a proven, well-understood approach. While operators that roll out Ethernet backhaul for the first time might well continue to rely on TDM timing initially, operators that are already seeing very large volumes of mobile data traffic on their networks and already have Ethernet backhaul in deployment know that one of the next steps in the evolution of their backhaul network roadmaps needs to be the elimination of TDM from the backhaul network altogether, which breaks the synchronization chain and impacts network timing. There are several reasons for that. As mobile data traffic grows and Ethernet backhaul matures, leaving TDM at the site becomes an unnecessary cost from several perspectives: Where the backhaul is leased, leaving an E1 at the cell site can typically cost anywhere from $150 to $500 per month in recurring opex charges. Across 5,000 sites, that's $7.5 million to $25 million in annual opex. Even where the backhaul is self-provided, such as via hybrid microwave solutions that support both TDM and Ethernet, TDM makes much less efficient use of microwave radio spectrum than Ethernet in a data-intensive application environment. There is also an operational cost associated with maintaining dual equipment types, operating procedures and maintenance practices for TDM and Ethernet at the cell site. These vary according to the local market, the operator's chosen architecture and the particular equipment types, but Heavy Reading knows of operators that have estimated running TDM and Ethernet service in parallel adds as much as 25 percent to their maintenance costs, as compared with a pure Ethernet environment. TDM provides frequency synchronization consistent with 3GPP and 3GPP2 requirements for the major 2G and 3G cellular standards, but it cannot provide time/phase synchronization of the kind that is mandatory for the TD-LTE variant of LTE, and will also be required for MBMS-SFN and Network MIMO features that operators may need for the FDD variant of LTE. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 3

4 TDM Timing Alternatives: A Fragmented Market As operators have trialed and begun deploying Ethernet backhaul over the last four or five years, until now a key factor that has prevented most of them from giving up TDM-based synchronization and migrating to alternative standards has been the fragmentation in the take-up of alternative proprietary and standardsbased synchronization solutions. Figure 1: The Three Main Standards-Based Approaches to Network Timing STANDARD MAIN STRENGTHS MAIN WEAKNESSES GPS (Satellite) IEEE 1588v2 (Packet Protocol) Synchronous Ethernet (L1 PHY) Source: Heavy Reading Already deployed widely throughout the U.S. for E-911 and in CDMA networks worldwide. Supports all timing techniques frequency and time/phase. Supports all timing techniques frequency and time/phase. Widely supported by equipment vendors. Network-wide upgrades not required. Conforms to the operator s familiar way of implementing L1 synchronization. Supported widely by equipment vendors. Relatively few deployments by GSM and W-CDMA operators. Resistance outside the U.S., as it is U.S. government-owned. No good for indoor deployments. Antennas can t "see" GPS satellites. New packet protocol, which requires new skill sets in the transport team to deploy and operate. In trials dating back to 2006, many operators have found 1588 operationally challenging to implement with the right level of performance. Only supports frequency timing. No roadmap for time/phase timing. Extensive network equipment upgrades required. Among the minority of the world's mobile operators that already have GPS at their cell sites, some are using its inherent timing capabilities rather than investing in a dedicated new synchronization solution. Operators using Ethernet over copper have trialed and deployed the NTR standard, which is specific to copper deployments. Two relatively new standards the ITU's Synchronous Ethernet and the IEEE's 1588v2 Precision Time Protocol (PTP) have been widely trialed and are also starting to be deployed in volume by some leading operators. And there have also been a large variety of proprietary solutions trialed and deployed, some of which are proprietary implementations of Adaptive Clock Recovery (ACR), some of which are based on other proprietary techniques. Of the variety of different timing solutions that have been deployed, only GPS, Synchronous Ethernet and IEEE 1588v2 are standards-based solutions that are widely applicable across multiple Ethernet backhaul environments. And because most operators don't want to get locked into a vendor's proprietary implementation, it is mainly these three standards that the operators are looking to for longterm synchronization of the mobile network. When it comes to network synchronization without TDM, it is clear that there is no "one size fits all." A greenfield operator that determines it has no long-term need for time/phase synchronization can manage perfectly well just with Synchronous Ethernet, because these days all of the new equipment deployed in the relevant HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 4

5 network segment is sure to be Synchronous Ethernet-enabled. Similarly, a CDMA network evolving to LTE can leverage just GPS at most of its sites for all of its current synchronization requirements, though other solutions will need to be found for the increasing number of in-building small cells that are expected to be required and for which the line of sight of the GPS antenna to the satellite is often obscured. Not only is "one size fits all" the wrong approach from one operator to the next, many operators are increasingly realizing that their long-term interests may well be best served by deploying two of the new synchronization solutions together, rather than being reliant on just one. One example might be to deploy Synchronous Ethernet for frequency synchronization together with 1588v2 for time/phase. In this scenario, Synchronous Ethernet helps lock down the frequency synchronization, thereby making the very challenging task of achieving accurate time and phase synchronization a bit easier for the 1588v2 protocol. Another scenario might be for an operator to use GPS at macro and micro cell sites but 1588v2 for small cells, with both protocols supporting frequency, time and phase. Increasingly, operators think in terms of a so-called "Any 2 Will Do" approach to synchronization standards. The Unique Value Proposition of IEEE v2 is well suited for deployment in conjunction with other standards according to several deployment scenarios and offers some key advantages over other standards. In addition to supporting frequency synchronization, 1588v2 natively supports time, as it is a "time exchange protocol," and there is a roadmap for supporting time and phase for telecom applications. Further, leading vendors have already started implementing time/phase as well as frequency synchronization in their 1588v2 implementations. Indeed the European Advanced Networking Test Center (EANTC) testified that in its February 2011 Advanced MPLS and Carrier Ethernet Multi-Vendor Interoperability tests, a number of vendors' backhaul equipment passed its interoperability tests for 1588v2 phase synchronization. In stark contrast, there are no industry plans to develop a phase synchronization roadmap for the Synchronous Ethernet standard. And there are no political barriers to the adoption of 1588v2 by operators around the world, whereas there are with GPS because of its ownership by the U.S. government. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 5

6 1588 in the Field: Initial Backhaul Deployments Many of the early deployments of the 1588 standard in Ethernet backhaul, going back to 2006, proved incredibly challenging. Up to now, and with few exceptions, the results from most trials of 1588 with Ethernet backhaul were not good enough to allow operators to move forward with removing their E1s or DS1 from the cell site and move forward with large-scale commercial 1588 deployments. The primary reason for this has been that operators have often found that the performance of either the 1588 protocol implementation or the underlying Ethernet backhaul network itself was too error-prone to allow the protocol to perform as required specifically against the critical network key performance indicators (KPIs) of packet delay, packet loss and in particular packet delay variation (PDV). The trick to getting the 1588 standard to perform as mobile operators require lies in ensuring that the 1588 packets are all transmitted at fairly regular intervals and also received at fairly regular intervals across the network. Several factors have conspired to cause a level of variability in 1588 performance that has often fallen short of operators' expectations Several factors have conspired to cause a level of variability in 1588 performance that has often fallen short of operators' expectations. It is important to consider that in the timeframe the deployment of the underlying Carrier Ethernet backhaul in mobile networks was itself in its infancy, with all of the teething problems that went with that. Backhaul equipment vendors implementing 1588 were also building their own implementation of what was for them a new standard. They often built their implementations in software, and often in different ways relative to other vendors. Different vendors also implemented 1588 differently relative to how they aligned with key ITU standards such as ITU-T Recommendation G.8261, "Timing and Synchronization Aspects in Packet Networks," as well as 3GPP , which specifies that the accuracy for frequency timing delivered at the cellular base station should be at least 50 parts per billion. Variability in critical delay-oriented metrics has also been compounded by the very different topologies and numbers of hops in different parts of different operators' backhaul networks. The effect of these variations in network design can be particularly noticeable in microwave backhaul networks, not only because of the variation in topology and hop count across different networks, but also due to the additional variability introduced by each vendor's proprietary algorithms for dealing with network congestion. Microwave links often cover greater distances over the "last mile" of the backhaul than copper- or fiber-served sites. The rest of this White Paper focuses on charting the evolution of the 1588 standard as a value proposition for mobile networks over the last four or five years. As this White Paper will show, what began as a complex, cutting-edge proposition with little or no supporting ecosystem of carrier-grade features or implementation guidelines has now evolved into an increasingly plug-and-play proposition for synchronization that operators can feel confident about deploying in volume. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 6

7 1588v2: The Telecom Profile It is important to note that any 1588 deployments that took place prior to 2008 were done with a basic version of the 1588 standard that was not yet optimized for telecom network applications. IEEE , also known as 1588v2, was released in Compared with 1588v1, 1588v2 supports higher accuracy, optional shorter frames, varied update rates, rapid reconfiguration after network changes, fault tolerance and the introduction of transparent clocks. All of these have rendered the standard much more suitable for implementation in mobile backhaul and other telecom network applications. The Telecom Profile defines additional elements to support 1588v2 frequency synchronization over telecom networks Most importantly, IEEE 1588v2 introduces the concept of a profile, which defines a set of parameters such that different devices can smoothly interoperate with each other. The intention is that specific industries like telecommunications, smart grid, manufacturing, digital audio and video broadcasting would develop their own profiles of 1588 for implementation in their own unique environments. In 2008, the ITU began the work of defining the Telecom Profile. Although the industry has had pretty clear visibility of what the final profile would look like since 2009, it wasn't until October 2010 that the Telecom Profile (ITU-T G8265.1) was formally ratified by the ITU. The Telecom Profile defines additional specific elements to support 1588v2 frequency synchronization over telecom networks. This includes using IPv4, unicast, signalling to support client negotiation, end-to-end PTP with no on-path support, and a redundancy model based on an alternate grandmaster. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 7

8 A Mature Carrier Networking Equipment Market The early waves of 1588 trials in Ethernet backhaul networks were implemented by operations teams in the operators that were often deploying Carrier Ethernet networks for the first time using Carrier Ethernet and synchronization equipment that was relatively new and unproven in any telecom network, let alone in the unique environment of a mobile network. Unsurprisingly, this created a lot of challenges in terms of getting the 1588 protocol to perform well. The early waves of 1588 deployments were carried out using vendors' own implementations of the protocol, with variable chip/packet processing rates, all of which made it very challenging to get the performance right, particularly in a multi-vendor environment. There is a lot of progress to report on this front. The 1588v2 standard never did, and still does not, mandate a specific synchronization rate (i.e., number of synchronization messages sent per second). An issue with the early deployments was that the 1588v2 grandmaster synchronization clocks were typically fixed in the synchronization rates that they supported. On the client or slave side, however, vendors were implementing a variety of synchronization rates. With roadmap investments on the grandmaster side, these products can now support nearly any synchronization rate from nearly any slave or client device. Whereas early 1588 trials and many subsequent 1588v2 trials from 2008 onward relied on vendors' own implementations of 1588v2, often in software, synchronization protocols such as 1588v2 and Synchronous Ethernet are now embedded in a variety of commercial third-party silicon from a variety of silicon vendors. This has had a tremendous impact on the cost, performance and time to market associated with the 1588v2 implementations of leading infrastructure vendors. Interoperability Before work on 1588v2 got underway, 1588 interoperability between vendors was virtually impossible. The start of the work on 1588v2 in 2008 was supported and continues to be supported today and going forward by large-scale bilateral and multi-vendor interoperability activities to ensure that different vendor implementations can interoperate. The level of vendor interoperability as well as the independent validation of vendor proof-points relating to 1588 today is unrecognizable from what it was even two or three years ago. For example, in addition to carrying out early validation of interoperability of some leading vendors' initial implementations of 1588v2 for phase synchronization, EANTC's Interoperability event in February 2011 also introduced testing of the new Boundary Clock and Transparent Clock features of 1588v2 for the first time. In its report following the February 2011 event, EANTC confirmed that several vendor pairings were able to interoperate successfully their Boundary Clocks as well as their Transparent Clocks. And the ITU is also working on performance specifications for both these new clock types. The 1588v2 Conformity Alliance At the end of 2010 the formation of the IEEE 1588 Conformity Alliance and the IEEE 1588 Conformity Assessment Program (ICAP) were announced. Drawing on the respective requirements of the IEEE 1588v2 standard and ITU-T Recommendations on timing and synchronization such as the Telecom Profile (ITU-T G ), the ICAP is designing a test suite for conformity testing of 1588 implementations. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 8

9 The test suite is being developed by the IEEE 1588 Conformity Alliance's Committee of Experts, whereas testing and certification activities will be managed by Iometrix, the Alliance's authorized test lab. The IEEE Test Suite for Frequency Synchronization in Telecommunications Networks focuses on conformity testing of packet master clocks and packet slave clocks for end-to-end distribution of frequency in packet networks. As of this writing, the certification program is due to go live in mid Consistent with the standard's development roadmap to also support time and phase synchronization, the Conformity Alliance is also targeting new certifications as follow-on work, focusing specifically on frequency synchronization. The IEEE 1588 Conformity Assessment Program could serve as a template for similar activities dedicated to conformity verification of Synchronous Ethernet systems. Figure 2: ICAP s Certification Test Development Suite for 1588v2 Conformance Source: IEEE 1588 Conformity Alliance The MEF Implementation Agreement & QoS in the Backhaul Network The first waves of 1588 trials were carried out on Ethernet backhaul networks that were themselves being trialed by the operators for the first time. Carrier Ethernet itself was nowhere near as mature as it is today. And to make matters worse, the mobile network was invariably behind the curve in rolling out Carrier Ethernet as compared with the wireline network. There are several examples of key advances that have rendered the mobile network a much more hospitable environment for enabling the 1588v2 protocol to perform as it should. In February 2009, the Metro Ethernet Forum published its first "Carrier Ethernet for Mobile Backhaul Implementation Agreement." This was a key HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 9

10 step forward because, recognizing the unique requirements of the mobile backhaul environment, such as the requirement for handoff between base stations, the Implementation Agreement laid down detailed guidelines for operators on how to design and engineer their Ethernet backhaul networks for maximum performance. It covers four major use cases that guide the service provider in setting performance and quality-of-service (QoS) parameters, synchronization options, recovery and protection mechanisms, as well as Ethernet OAM services. The MEF's Implementation Agreement was one of the key contributions to the marked improvements in the QoS environment, as it affects 1588v2's performance in the mobile network. Although there is still some way to go before it is perfected across the mobile network, there has been very strong progress made in aligning QoS requirements across different RAN, backhaul and core elements of the mobile network, as well as between different 3GPP, IP/MPLS and Layer 2 Carrier Ethernet QoS environments. This progress makes 1588v2 implementation much easier than it was a few years ago, for example by enabling the 1588v2 packets to be isolated into their own dedicated QoS class and given absolute priority across the network. This enables them to bypass all the standard queuing mechanisms in the switches, routers or microwave nodes in the network. Before this capability came within reach of the operators, in trials 1588v2 packets inevitably proved just as vulnerable as any other packet to jitter, delay and delay variation. Hence the protocol often failed to meet the operator's synchronization requirements. Consider also that when 1588 was first trialed, it was nearly always in conjunction with pseudowire capabilities for emulating legacy TDM and ATM services over packet. Again, at that time pseudowire technology was in its infancy and was particularly challenging to get right in the last mile of the backhaul, in part because of the overhead generated in a relatively low bandwidth environment. Great strides have also been made in optimizing pseudowire deployments and aligning implementations between different vendors to reduce the impact of pseudowire emulation on the performance of 1588v2 packets, as well as the variation in impact from one vendor to the next. In addition, with the large-scale deployment of new base stations with native Ethernet interfaces, more and more packet backhaul deployments don't require pseudowires in the last mile or access part of the backhaul at all, further simplifying the environment for synchronization. The Syncworld Ecosystem Program for Pre-Integrated Solutions In March 2011, Symmetricom, the world's largest supplier of 1588v2 synchronization clocks, announced the creation of its Syncworld Ecosystem Program. Together with more than 14 vendor partners, including Cisco Systems, Aviat Networks and Telco Systems, Syncworld is a key industry initiative whereby Symmetricom and its partners are providing very detailed interoperability and implementation documentation for operators to ensure seamless deployment of precise timing and synchronization across the entire network. These blueprints specify, in a lot of detail, guidelines on exactly how each specific vendor's equipment can be optimally deployed with Symmetricom's 1588v2 clocks for optimal performance, taking into account known performance characteristics of each vendor's products. The detailed documentation assembled in Syncworld facilitates operator deployment. Whereas in the past each operator had to invest HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 10

11 a lot of time and money figuring out optimal configurations of equipment and clocks in their labs and in field trials and then go back time and again to realign them Syncworld enables each operator to avoid "reinventing the wheel" so far as many of the initial network design parameters are concerned for specific pairings of vendor products. For example, members of the SyncWorld program have performed the following interoperability tests: Aviat Networks conducted extensive 1588v2 over microwave testing, including a 10-hop microwave configuration with adaptive modulation, to validate both timing quality and provide guidelines for backhaul engineering. Cisco conducted testing to assure precise timing and synchronization carried across Cisco's Unified RAN Backhaul network solutions. Telco Systems conducted ecosystem interoperability testing verifying the resiliency of its demarcation device over different microwave links (including E-Band) and fiber topologies, including 1588 boundary clock testing done in the last EANTC event. Mature Test Suites Mature testing solutions, enabling the performance of the 1588v2 implementation in specific network configurations to be subjected to simulated stress tests and network failure scenarios, have come onto the market only in the last year or two. The maturity of present-day testing suites now gives operators more confidence that they can scale their networks up roll out 1588v2 to dozens, hundreds or thousands of cell sites because they now have a much better idea of how their network will respond in multiple failure scenarios, and can design in redundancy and resiliency features accordingly. The absence of such commercially available tools had been a major barrier to many operators rolling out 1588v2 in volume. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 11

12 The Outlook of the Operators on IEEE 1588v2 Heavy Reading research has found new and compelling evidence that mobile operators throughout the world are increasingly readying themselves to give up on using DS1s or E1s for backhaul synchronization as they transition to packet backhaul. The same research also shows a marked preference for standardsbased solutions over proprietary ones; and that of all the relevant standards, and notwithstanding the challenges that most operators have had in getting it to work until recently, 1588v2 is clearly seeing the strongest demand from operators. The evidence is shown in Figure 3. This is taken from Heavy Reading's year-end 2009 and year-end 2010 annual surveys of qualified mobile operator respondents on their outlook for the evolution of the backhaul network. It shows the respondents answers to the exact same worded question asked in both the 2009 and 2010 surveys, namely: "On a scale of 1 to 5, where 5 is 'highly likely' and 1 is 'not likely at all,' please rate the likelihood that your company will adopt the following synchronization solutions as its preferred approach to provide synchronization of packet backhaul over the next three years." Figure 3: Average Operator Ranking (Out of 5) for Different Synchronization Solutions DECEMBER 2009 DECEMBER 2010 IEEE 1588v Synchronous Ethernet GPS Leaving a T1/E1 at the cell site NTP Adaptive Clock Recovery NTR Other proprietary solution Source: Heavy Reading Mobile Backhaul Survey, December 2010; 83 respondents In the 12 months between December 2009 and December 2010, using a DS1 or E1 at the cell site for packet backhaul synchronization slumped from being among the most popular options, at 3.4 out of 5; to lagging far behind the newer packetoriented standards, at 3.0 compared with 3.7 out of 5. The three primary standards-based approaches 1588v2, Synchronous Ethernet and GPS all scored significantly higher than any of the proprietary approaches, including ACR. And while the annual Heavy Reading backhaul survey series showed in December 2009 that operators believed Synchronous Ethernet would prove more relevant than 1588v2, that gap has now closed. In December 2010, the two standards were equally popular, at an average of 3.7 out of 5. Just as importantly, over the 12 months that separated the two surveys, the needle shifted more positively for 1588v2 than for Synchronous Ethernet or indeed any other synchronization solution. This shift in opinion suggests that the market's level of commitment to the 1588v2 standard is now steadily increasing. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 12

13 There are also some concrete commercial proof-points to support this proposition: A Symmetricom partner states that it had more than 40,000 cell sites in live service supported by 1588v2 with mobile operators around the world as of March BT still uses E1s for synchronization of the wholesale Ethernet backhaul service it delivers to U.K. mobile operators. But having trialed and discounted ACR solutions, BT is now preparing to roll out 1588v2 and Synchronous Ethernet as U.K. mobile operators evolve their HSPA and LTE networks. At least two of Vodafone's affiliate operators now have 1588v2 up and running in commercial service. Heavy Reading's Forecast for 1588v2 Adoption Heavy Reading's Ethernet Backhaul Quarterly Market Tracker has been tracking and forecasting the evolution of the global Ethernet backhaul market since November Our forecast for the adoption of various synchronization standards over the next four years is laid out in Figure 4 below. Figure 4: Heavy Reading's Forecast for Adoption of Preferred Synchronization Solutions Source: Heavy Reading's Ethernet Backhaul Quarterly Tracker, June 2011 Consistent with Heavy Reading's research, Figure 4 shows that leaving an E1 or DS1 at the cell site continues to be the most popular approach for synchronization in Ethernet Backhaul deployments, accounting for 43 percent of all live Ethernetserved cell sites at the end of Heavy Reading forecasts this legacy approach will continue to be the most popular approach through the end of 2012, at which HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 13

14 time it will still be providing synchronization for 35 percent of Ethernet backhaul deployments, and into Of the many fragmented and standards-based approaches, Heavy Reading expects IEEE 1588v2 to emerge as the most popular approach over the next 12 months, followed by GPS and Synchronous Ethernet. As the ability to support time/ phase synchronization is added to its current support of frequency synchronization, Heavy Reading expects IEEE 1588v2 to become the most widely deployed packet synchronization standard worldwide, providing synchronization to more than 1 million cell sites in live commercial service by the end of Consistent with the "Any 2 Will Do" strategy that Heavy Reading expects many operators to adopt, we are also forecasting strong growth in Synchronous Ethernet and GPS. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 14

15 Summary In order to grow the mobile broadband market beyond the early-adopter phase that we are seeing in most markets worldwide, mobile operators need a relentless focus on cost control. In the backhaul network, that doesn't just mean introducing lower-cost Ethernet service. At some point, depending on the operator's migration strategy, it also requires getting rid of the cost overhead associated with leaving TDM at the cell site for network synchronization. The IEEE's 1588v2 standard is emerging as the preferred new synchronization standard because of its extensive support among operators and equipment vendors, and because of its support not just for frequency synchronization but also for time and phase synchronization. There is no doubt that most early deployments of 1588 failed to meet operators' expectations with respect to performance. This was due to the initial lack of optimization of the protocol for mobile backhaul applications, as well as topology, dimensioning and performance issues within the underlying Ethernet backhaul networks themselves, as well as the absence of clear operating and dimensioning guidelines and testing capabilities. In the last two or three years, extensive industry cooperation among operators, vendors and standards bodies has created a powerful ecosystem. This is working to ensure that the past performance issues with 1588 are now being successfully tackled, enabling the standard to become an increasingly plug-and-play solution for mobile backhaul synchronization. HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 15

16 About This Paper About the Author Patrick Donegan Senior Analyst, Wireless, Heavy Reading Donegan has been a telecom market journalist, analyst, and strategist for 20 years. He joined Heavy Reading from Nortel, having spent five years as a senior manager of strategic planning for that company's wireless business spanning GSM, CDMA, UMTS, WiMax, and other wireless technologies. Prior to Nortel, he spent two years in business research for Motorola's Corporate Strategy Office in EMEA and two years as a wireless analyst for the Yankee Group. At Heavy Reading, Donegan has focused on next-generation mobile network issues. Donegan is based in the U.K. and can be reached at donegan@heavyreading.com. Original Research This Heavy Reading White Paper was commissioned by Aviat Networks, Cisco Systems, Symmetricom and Telco Systems, but is based on independent research. The research and opinions expressed in the report are those of Heavy Reading. Heavy Reading's annual mobile backhaul survey referenced in this White Paper was carried out in December qualified respondents in operators with mobile networks responded, with a broad distribution of responses from across the world's major geographical regions. About Heavy Reading Heavy Reading, part of the Light Reading Communications Network, is an independent market research organization offering quantitative analysis of telecom technology to carriers, service providers, and vendors. Our remit is to provide the comprehensive competitive analysis needed today for the deployment of profitable networks based on next-generation hardware and software. Heavy Reading 240 West 35th Street, 8th Floor New York, NY USA Phone: HEAVY READING JUNE 2011 WHITE PAPER IEEE 1588 TIMING FOR MOBILE BACKHAUL: THE ROAD TO PLUG & PLAY 16