Ethernet Backhaul Battle Brews

As wireless carriers get serious about deploying Ethernet backhaul, equipment vendors will be honing their technological arguments to try to prove their gear does the job.

The result could be a long debate, because, while nearly every Ethernet vendor says it does wireless backhaul, some are saying the job is harder than it looks.

At next week's Ethernet Expo 2007, to be held in New York City, wireless backhaul promises to be one of those topics you can't escape from. Two of the panel sessions, one each on Monday and Tuesday, will specifically discuss backhaul, and the topic is likely to emerge as vendors discuss areas such as Ethernet access gear. (See Access Heats Up for Ethernet and An Ethernet Expo Rundown.)

The general theme behind backhaul is this: Bandwidth demand could shoot upwards as mobile data services come into play, overwhelming the leased T1s and E1s currently used for backhauling voice.

Equipment vendors are hoping this pushes mobile providers to consider Ethernet for backhaul. "It's really compelling from a short-term perspective to keep adding T1s or E1s, but you can never add enough to fix the data problem," says Craig Easley, vice president of marketing for Actelis Networks Inc.

The catch is that the Ethernet argument will fall flat if mobile bandwidth demand doesn't surge for some reason. In other words, Ethernet becomes a winner only if cell sites need a lot of T1/E1s. "Most mobile and wireline carriers report that the business case becomes valid at around the third or fourth T1/E1," Heavy Reading analyst Patrick Donegan wrote in a May report, "Ethernet Backhaul: Mobile Operator Strategies & Market Opportunities."

The other factor noted in the chart above is the distance between the base transceiver station (BTS) and radio network controller (RNC), since T1/E1s often are leased by the mile.

Once a carrier decides to go with Ethernet backhaul, any old Ethernet box won't do. Vendors chasing this market in the coming months will have to answer to several technological points.

  • Timing: T1s and E1s carry a clock signal that Ethernet lacks. Without it, performance of voice and video traffic could suffer. Plenty of standards are being kicked around here, and companies like Brilliant Telecommunications Inc. and Symmetricom Inc. (Nasdaq: SYMM) are chipping in with timing products. (See Luminous Crew Gets Brilliant.)

    Some proposed answers are based on using a GPS clock, but that's a $50,000 expense, Easley says. Others rely on standards -- Institute of Electrical and Electronics Engineers Inc. (IEEE) 1588v2 or International Telecommunication Union (ITU) G.8261. No option has emerged as an obvious winner, although 1588v2 seems to be a favorite.

  • Reliability: As with carrier Ethernet services, wireless backhaul has to be as sturdy as a T1. Ethernet demarcation boxes, already a hot topic in carrier Ethernet, are going to be key here, says Brian McCann, chief marketing and strategy officer for ADVA Optical Networking .

    That's because cell sites are so spread out. Operators will want ways of pinpointing the location of a line break, for instance, without sending trucks all over the place. They'll also want a means of monitoring network performance to compare it to service level agreements, McCann says.

  • The outdoors: In many cases, McCann notes, cell sites will require temperature-hardened gear that can survive a long, lonely life inside a telecom hut.

  • 800-pound gorillas: As with any Ethernet market, Cisco Systems Inc. (Nasdaq: CSCO) could cause trouble for smaller companies. A Cisco offering "may be more expensive and more complicated, but it's comfortable," McCann says.

    It's worth noting Alcatel-Lucent (NYSE: ALU) wants a stake in this market, too. And even without those two giants, startups in this space would have to contend with offerings from Adtran Inc. (Nasdaq: ADTN) and Tellabs Inc. (Nasdaq: TLAB; Frankfurt: BTLA). (See AlcaLu Targets Wireless Backhaul and Tellabs Snares Euro Backhaul Deal.)

    — Craig Matsumoto, Senior Editor, Light Reading

    Interested in learning more on this topic? Then come to Ethernet Expo 2007, a conference and exhibition that will explore major carrier Ethernet networking and service trends. To be staged in New York City, October 15-17, admission is free for attendees meeting our prequalification criteria. For more information, or to register, click here.

  • Page 1 / 2   >   >>
    bgopi 12/5/2012 | 3:46:24 PM
    re: Ethernet Backhaul Battle Brews Can anyone tell me how to simulate the GPS source for testing purposes?

    Also, any familiar ways to verify the timing when TDM networks replaces with PSN backhauls? Like , i may want to run a live voice or video to test the quality and loss of data, but are there any tools for that.

    I have used VLC sometime back on DSLAM, but, not sure, if that can be used in TDM-Eth networks.
    OldPOTS 12/5/2012 | 3:00:51 PM
    re: Ethernet Backhaul Battle Brews The problem is how do you connect/transport traffic from all those diverse BTSs in a hub and spoke architecture to a RNC. Many of whom require less BW than a T1/E1 over copper wire or microwave systems. Maybe when FIOS completely covers enough of an RNC area it might be viable.
    Rjm this might need for a local government network.

    The question for Ethernet is what does one use to carry the ethernet. Cost of equipment is relatively small verses transport cost (many RFP responses). Yes this transport cost is a significant cost to mobile/wireless carriers. Also most BTSs and RNCs have methods to compress/reduce voice BW. The problem appears to be for the very few BTS sites that have enough traffic (probably heavy data/video) that require multiple T1/E1s.

    Back to Ethernet (protocol) over T1/E1 (transport).
    The cost for equipment becomes additional cost.
    Also a T1 can be used to supply network timing to the Ethernet (switch) as it flows outward from the expensive CO clock.


    PS Asked how to solve this problem every few months, during many years when mobile/wireless was available before I retired.
    opticalwatcher 12/5/2012 | 3:00:51 PM
    re: Ethernet Backhaul Battle Brews I've seen GPS clocks for less than $1K, such as:
    But I'm no wireless backhaul expert, so maybe I'm leaving out some other required expenses to make it all work when using Ethernet instead of T1/E1 links.
    macster 12/5/2012 | 3:00:50 PM
    re: Ethernet Backhaul Battle Brews When TDM backhaul is replaced by Ethernet backhaul, base stations get isolated (sync-wise). There are ways to restore sync; embedded stand-alone clocks, packet-based sync protocols and GPS re-timers.

    Regarding PSN sync, there are two standards, IEEE 1588 (V1 and V2) and ITU G.8261 (Synchronous Ethernet).

    The IEEE 1588 solution requires that all NEs support IEEE 1588 (PTP and bi-direction), which is a GbigG upgrade. Some say SAToP and CESoPSN both uni-directional) are sufficient. But IEEE 1588 gives Gtime of dayG, which is useful for billing, etc.

    Although packet networks are meant drive convergence, I believe that synchronisation should be de-coupled. UMTS Node BGs, e.g. will come with embedded clocks which meet the 50 ppb requirement.
    macster 12/5/2012 | 3:00:50 PM
    re: Ethernet Backhaul Battle Brews There are many way to connect a hub-and-spoke architecture - GPON gives up to 32' branches', PMP radio which can do stat-mux, etc.

    One can use coppper (EFM), fibre and also radio to carry ethernet.

    Agreed that Abis and Iub optimisation reduce voice bandwidth, but we are talking about data. The article mentions the big 'IF' in terms of bandwidth growth. But given that standards are ever increasing uplink/downlink speeds, e.g. 100M/50M for LTE, true mobility (not nomadic) will also be a driver.
    rt369 12/5/2012 | 3:00:49 PM
    re: Ethernet Backhaul Battle Brews Good previous post.

    Wireless data today: 3mpbs
    T1 Cost: $200/mbps

    Wireless data under 4G: 50mbps+
    Target transport cost: $20/mbps for pipes of 100mbps.

    And fisrt provider to the cell with fiber wins (at least for the first 60 month term)

    MW can be useful for tail ckts but it will be necessary to dump it onto fiber eventually.
    macster 12/5/2012 | 3:00:48 PM
    re: Ethernet Backhaul Battle Brews There are new challenges raised by packet backhaul.

    Carriers that provide leased lines also provide an SLA level. TDM demarc devices are well understood. Indeed, EFM over copper can provide 45Mb, but in order to provide QoS, one needs an Ethernet demarc (802.1aj) device. It all comes down to how determined and willing carriers are to overhaul their access networks, what with leased lines still providing a very nice revenue stream. This in turn depends on market demand and product availability, e.g. a simple 1-port etherjack kit.

    PMP radio solutions that do stat-mux is a real boon - this will dump onto a fibre core.

    This core does not need to be MPLS but can be Carrier Ethernet Transport. There is a difference between mobile packet backhaul and backbone!

    Sync is an issue, but there are solutions - saying that, I believe sync should be de-coupled from backhaul.

    OldPOTS 12/5/2012 | 3:00:41 PM
    re: Ethernet Backhaul Battle Brews This is the wireless teenager (growing fast) in the company telling the wireline parent to modernize.

    The parent doesn't understand why it should invest more money to replace a solution that still works.
    Besides Ethernet hasn't been as cheap as first perceived, so we just don't yet trust it for a growing revenue stream.

    macster 12/5/2012 | 3:00:41 PM
    re: Ethernet Backhaul Battle Brews This is the wireless teenager (growing fast) in the company telling the wireline parent to modernize.

    The parent doesn't understand why it should invest more money to replace a solution that still works.
    Besides Ethernet hasn't been as cheap as first perceived, so we just don't yet trust it for a growing revenue stream.



    OP, I seriously don't get your posts :-( It's probably just me, but I just can't make sense of it. Sorry...

    jliu229 12/5/2012 | 3:00:37 PM
    re: Ethernet Backhaul Battle Brews There are three kinds of Network Synchronizations: Clock Frequency Synchronization, Clock Phase Synchronization and Time Synchronization. Clock frequency synchronization means all Base Stations and Base station controllers work at the same frequency, and it comes from two sides: one is radio interface and the other is TDM-based transport network (T1/E1/SDH/SONET). 2G/3G networks all require +/-50ppb clock accuracy among base stations . Usually the clock is recovered from T1/E1 line at BTS. When the T1/E1 is replaced with regular Ethernet, it is almost impossible to achieve the same kind of frequency accuracy due to packet delay variance resulting from packet size variance and packet queuing and scheduling, but may be OK with P2P ethernet link.

    In CDMAOne and CDMA2000 networks, base-stations are identified by different time offsets of the same scrambling code, thus the SYNC signals to all Base stations must be phase aligned (each Base Station will make adjustment per provisioned amount of time). In addition, to minimize the next cell searching effort for handoff, the MS needs to know the next cells' SYNC position as accurate as possible. The T1/E1 Frame SYNC signal cannot be used as it has nothing to do with the distance from the Base Station Controller. When T1/E1 line is repalced with Ethernet, it becomes worse due to Packet Delay variance. WCDMA uses unique scrambling code to identify each Base station thus no need for Clock Phase Synchronization among Base Stations. TD-SCDMA use the same scheme as WCDMA, but it requires all Base Stations Clock Phase aligned to minimize searching effort for the next cell.

    The time-of-date (such as 2007:10:17:16:32:400)synchronization means everyone use the same time reference, it is mainly for accounting and reporting purpose.
    Page 1 / 2   >   >>
    Sign In