Article Info

The Wireless Backhaul Opportunity

The best session of the Lightreading Ethernet Conference covered Wireless Base Station Backhaul. Patrick Donegan of Heavyreading and the panelists presented cohesive data and their take on which way the market would head. I gained a new perspective on the opportunity wireless backhaul presents.

The fundamental problem wireless carriers face is the underlying shift from voice dominated to data dominated traffic. Voice is growing linearly, while data is likely to grow exponentially. If leased copper T1’s are used for backhaul, their backhaul costs will scale linearly as capacity is added.

Understanding the demographics of the problem are crucial to uncovering the opportunities. HeavyReading presented the following:

  • There are approximately 2.2M wireless base station sites worldwide.
    • Asia 49% (1.1M)
    • North America 11% (250k)
    • Central/Latin America 5% (125k)
    • Europe 29% (575k)
    • Mid East/Africa 7% (150k)
  • 50% of global base stations use microwave for wireless backhaul. The remaining half are evenly split between fiber and copper.

Mashing up additional data from the panelists along with my own opinions yields the following:

  • Of the 550k sites using copper backhaul, virtually all of them use T1/E1 PDH connections.
  • China Mobile and China Unicom have about 500k base stations, the majority of which are served by fiber (not necessarily connected)
  • Fiber reaches only 20% of the 250k North American base stations. Microwave is virtually nonexistent. 90% of NA base stations are copper connected, 10% are fiber connected. All copper is TDM T1 today.
  • The opposite is true in Europe/MidEast/Africa where 60% of connections are microwave, 25% are fiber, and only 15% are copper. Stiff copper E1 pricing in Europe forced an evolution to microwave early on. 
  • Between 20% and 40% of wireless carrier opex is backhaul, depending on how much of the backhaul infrastructure is owned by the carrier. Verizon has low costs for copper fed base stations where they provide the T1s.
  • Sprint has very high cost backhaul. Moving their backhaul growth to Wimax yields a large cost savings.

Carriers don’t like copper T1s because of the high cost per bit but they will trust them with their lives. Given traffic growth is ‘best-effort’ data, carriers are looking at running parallel networks where voice backhaul is on reliable T1’s, and data is moved over to Ethernet. This caps their high cost T1 copper expenditures and moves the growth to a lower cost backhaul method.

The Opportunity

Europe does not appear poised to change in the near term, other than a move from TDM (E1) based microwave to Ethernet based microwave. One company I plan to look at is Ceragon Networks (CRNT). They would appear to benefit from this trend though the company if far from ‘undiscovered’. They OEM their products through Nokia. Other suggestions are welcome.

North America is much more interesting. The current reliance on expensive copper T1’s as a backhaul technology will simply have to end, the only question is how it changes.

Symmetric DSL backhaul is one solution, but recent forbearance rulings by the FCC force me to question whether ILECs will offer this service at a competitive price.  Hatteras, Actelis and others who are pure-play SDSL vendors count wireless backhaul as a key business driver. I think the FCC just drove a stake through their heart. Comments welcome.

I have heard that a major US wireless carrier (Verizon, I believe) is moving to microwave based backhaul in the unlicensed 5GHz spectrum. I don’t know who is providing the equipment, perhaps Harris-Stratex (HSTX) ? Fibertower (FTWR) provides T1 equivalent backhaul using 30GHz microwave but looks like a classic infrastructure cash sinkhole. Perhaps it will be saved by a combination of monopolized DSL and an urgent need for cheap data backhaul.

Will the big tower providers, Crown Castle (CCI) and American Tower (AMT), evolve to provide Ethernet ISP services in addition to hard infrastructure? It would make sense to me that either of these companies could be in the backhaul business.

From a silicon perspective, the opportunity size is simply not that big. There seems to be a legitimate hesitancy on the part of carriers to move voice T1’s to circuit emulation because the risks outweigh the potential rewards. This puts a dark cloud over circuit emulation as a big chip market. Ethernet over PDH could have more legs, particularly if the ILECs price T1’s to margin to fend off competing technologies. Unfortunately there are only 400k copper sites in North America; assuming a few chips per site and $50 a chip yields <$100m total market over many years. Silicon vendors who are banking on wireless backhaul as their big market are likely to be disappointed.


Europe and Asia appear to be poised for evolutionary change with high bandwidth Ethernet based microwave and straight fiber, while North America will see a revolutionary change away from TDM copper to microwave and potentially symmetric DSL. The opportunity appears to be on the services side, and perhaps a focused equipment provider. The silicon opportunity here is relatively small.


Comments are disallowed for this post.

  1. Interesting. Also gives hard numbers on why US cell coverage sucks compared to Europe.

    Posted by Zed | October 25, 2007, 2:08 AM
  2. From the product point of view, BridgeWave (US, privately owned) and DragonWave (public, Canadian and AIM quoted) seemed to have other good high-bandwidth (> 100 Mb/s) wireless point to point link products that could be used instead of Ceragon’s. You can see others digging through this vendor list: No idea about market shares or valuations, though.

    Posted by Carlos | October 25, 2007, 3:16 AM
  3. Andrew,

    The first two bullets are an accurate representation of the data I presented. Thank you for that. The remaining bullets are yoiur own conjecture and do not accurately reflect my view of the market.

    Please re-jig nyour piece so as to clearly separate the two eleemnts of the article out rather than having them appear as a continuous piece.
    Thank you.

    Patrick Donegan

    Posted by Patrick Donegan | October 25, 2007, 4:18 AM
  4. Patrick:

    The bullets are a summary of what was presented by the panelists and yourself. I re-read it and am not sure how to improve on what is there. I don’t think it leaves the impression these were your opinions.

    The first two bullets were your data. I also remember you highlighting the China Telecom and Unicom facts.

    Feel free to comment with any points you disagree on. I meant no harm – it was a very good panel.

    Posted by Andrew Schmitt | October 25, 2007, 7:18 AM
  5. FTWR certainly has to manage their growth smartly or your sinkhole comments will come true

    Posted by JC | October 25, 2007, 9:17 AM
  6. You always want to be the _last_ investor in any big capex endeavor. Virtually all of them go belly up. Metricom, Panama Canal, Chunnel, Fiber boom of 2000 (pick your company here), Transcon Railroad, Iridium, Clearwire (yep), Nuclear power (so far),

    Exceptions: Erie Canal, FiOS (?), US Mobile Carriers (amazing huh?), FedEx, DK-Sweden Bridge.

    Management and a heavy dose of skepticism are the differentiators.

    Posted by Andrew Schmitt | October 25, 2007, 9:48 AM
  7. Glad you enjoyed the panel but the flow of your bullets is open to the interpretation that any or all of the datapoints are attributable to me when in fact they are not. Only the first two are. Seperating them out would make clear which are the Heavy Reading assumptions that you are reporting and which are the assumptions that you are generating yourself.

    Regarding China, I stated that China Mobile has 250,000 sites based on hard data. I said nothing about China Unicom. You are apparently confident to conclude that China has 500,000 cell sites. I’m not.


    Posted by Patrick Donegan | October 25, 2007, 10:13 AM
  8. Andrew,
    The Cablecos in the US are moving quickly into this area – they are focusing on stringing fiber to the cell towers. Telecoms Magazine had a seminar on Mobile Backhaul and there is a wealth of info on slides on the site there.

    Posted by Sajith | October 25, 2007, 10:18 AM
  9. I am keen to distance myself from most of your assumptions. In particular I am very keen that your readers should understand that I do not agree at all with your assumption that in EMEA, 25% of base stations connections are fiber, and only 15% are copper.

    Posted by Patrick Donegan | October 25, 2007, 10:41 AM
  10. Got it. I’ll contact Actelis, which is the source of the information.

    I have made your requested changes to the article.

    Thanks for your comments.

    Posted by Andrew Schmitt | October 25, 2007, 10:56 AM
  11. Good article and interesting insight into the regional variations driving backhaul technology choice. Your assumption of 500k cellsites in China seems reasonable at first glance to me (with 500M subscribers, a ratio of 1:1000 looks about right), but note that around 100M of those subscribers use PHS (officially not considered mobile technology because run by the fixed operator China Telecom). Since PHS will not be offering high data services comparable with 3G (it’s really an extended cordless phone standard more like DECT), this can be discounted for long term assessment of backhaul technology growth. If 2.2M basestations is correct, and there are 3Bn subscribers globally, then a ratio of 1400:1 is reasonable – with high 2G growth in India and other emerging markets, this ratio may increase in order to meet tight cost targets. However, the growth of 3G should allow this balance to be redressed in the longer term because of much higher capacity/throughput per cellsite. This assumption ignores potential disruptive solutions like femtocells and other wireless technologies.

    Posted by David Chambers | October 31, 2007, 6:26 AM
  12. Andrew-

    A few comments to add to your piece:

    “Fibertower (FTWR) provides T1 equivalent backhaul using 30GHz microwave but looks like a classic infrastructure cash sinkhole. Perhaps it will be saved by a combination of monopolized DSL and an urgent need for cheap data backhaul.

    Will the big tower providers, Crown Castle (CCI) and American Tower (AMT), evolve to provide Ethernet ISP services in addition to hard infrastructure? It would make sense to me that either of these companies could be in the backhaul business.”

    I have to agree with Patrick that your article is mis-leading and to a certain extent will lead folks knowledgable about the industry to conclude that you have not done too much homework to support your opinions. Right now Crown Castle (CCI) and American Tower (AMT) are in the backhaul business as they have a strategic partnership with Fibertower (FTWR) and are major financial backers of the nationwide wireless backhaul network.

    Your article which could have been informative in such an explosive growth industry only demonstrated your ability to make broad based assumptions and forecast industry happenings with little or no support.

    Next time do your homework and I will listen.


    Posted by Steve | November 3, 2007, 4:54 PM
  13. I don’t think AMT or CCI want to put themselves in direct competition with their customers so they sponsored FTWR. AMT and CCI are the largest tower operators by far in the US and they make can their towers more attractive by offering wireless backhaul. Lets say Sprint, ATT or DT had the choice of using an AMT tower which has both wireless or VZ copper backhaul or a Brand X tower which offers only VZ copper? Add in that wireless is less expensive and at least as reliable and the AMT and CCI towers with FTWR wireless backhaul become more attractive.

    FTWR really isn’t looking for the existing voice market to switch from T1 to wireless, they’re going after the data/video market which has vastly greater bandwidth demands than voice. Smart phones like IPhone, stuff like that.

    Interestingly, they’re teamed with VZ (and Qwest) for the new $20B federal Networx contract. Federal buildings are mandated to obtain redundant communications and wireless is an acceptable method.

    Posted by JC | November 4, 2007, 9:01 AM
  14. AMT and CCI have put themselves in direct competition with their customers because they can. They own the towers and essentially have a stranglehold on the wireless players because putting up new towers is extremely expensive and the municipalties/local governments are taking a stand. AMT and CCI have financially backed Fibertower in addition to Fibertower having the advantage of being the sole wireless backhaul provider on all of AMT’s and CCI’s towers. I do agree that AMT, CCI, and FTWR see the tidal wave of backhaul needs coming as a result of data. I think a 100KB file(Simple Digital Photo or Smartphone Web Page Equivalent) is equal to a 5 or 10 minute phone call. With wireless data demand growing 100 % YOY I see AMT, CCI, and FTWR in the catbirds seat when it comes to providing backhaul. Running more copper is pointless, running fiber to the cell site is very expensive and questionable long term, but backhauling back to the main fiber rings and the core through wireless microwave makes a lot of sense.

    Posted by Steve | November 4, 2007, 10:48 AM
  15. Steve:

    Not sure what you mean about not doing my homework considering I mentioned Fibertower right before AMT and CCI.

    Partnering with someone is not the same as doing.

    Also – no response from Actelis yet.

    Posted by Andrew Schmitt | November 5, 2007, 8:01 AM
  16. Question for Andrew and Patrick

    In your technical opinion, why the RAN optimization technologies from companies like Carrier Access ( and NMS Communications ( have not taken off? These technologies seem like cost efficient alternatives to expand bandwidth without employing additional T1s, however, carriers have been slow to adopt these technologies?

    Posted by michael goldman | November 16, 2007, 5:22 PM
  17. Responses to the article. My estimation that only 15% of the EMEA cell towers is based upon experience working with a number of CLECs and information received from PTTs who provide backhaul servicesand mobile operators who run their own backhaul networks. Because most of the backhaul networks at present are run by the mobile (wireless) operators the backhaul networks they built form themselves tend to be wireless as well, I think Patrick from HR and other analysts have put the number between 50-60%. Of the remaining 40% I estimated the split it 25-15 but it could be closer to 20-20.

    Commenting on Steve’s point about the futility of running new copper to the cell towers, I agree completely. But if you can get 4 times the bandwidth with Ethernet over Copper than you can with existing T1 services over the same copper, you don’t necessarily need more copper to deliver the capacity required by the next generation of mobile services.

    Posted by Craig Easley | December 3, 2007, 6:31 PM
  18. Patrick,

    I agree with the initial data you provided in your article, but I don’t agree with many of your assumptions regarding the North American market.

    First of all, mobile operators do not wish to own backhaul infrastructure. The real opportunity is for alternate fiber based service providers to add a fixed-wireless component to reach cell sites not cost effectively served by fiber. Mobile operators will add Ethernet services at cell site, but TDM services will continue to expand and be part of the package for many years. While MPLS and pseudowire look promising, mobile operators seek solutions that offer both TDM and Ethernet.

    Yes, Ceragon offers a very good solution; but so doesn’t Harris-Stratex, Nera, DragonWave, Microwave Networks, and others. I believe Ceragon, and the others, will enjoy the benefits of increased fix-wireless mobile backhaul.

    With regards to American Tower and Crowne Castle, they do not wish to become carrier’s carriers. They lease real estate, period. Fibertower is simply an investment that has synergy with the primary scope of business.

    The tower companies are not acting in a competitive manner.

    Mark Reynolds

    Posted by Mark Reynolds | January 11, 2008, 1:57 PM