Most cell sites have only one or two T1/E1 links, or less

One of the things that surprised me at eComm was the audience reaction to hearing that most cell sites have only one, or perhaps two, T1 or E1 links going to them.  eComm was a sophisticated audience but they weren't familiar with this kind of operational detail.

Of course, I've seen a lot of data on this particular subject as, until our recent sale of the AccessGate product line, NMS Communications was in the backhaul optimization business.

Net-net:  As spectral efficiency keeps improving with HSPA and so forth, backhaul is rapidly becoming the bottleneck. 

Cell sites are fixed locations.  The vast majority are in urban or suburban neighborhoods.  Their bandwidth requirements will continue to grow, indefinitely.  In any rational world, one would purchase dark fiber to most of these sites.  In the irrational real world that is the US today, dark fiber is available on long haul routes but is extremely hard to come by in populated areas where most cell sites are located.

The root problem is we have given Verizon and AT&T privileged access to the public right-of-way, without requiring them to sell dark fiber, or lease dark fiber, or offer any connectivity other than T1 service.

What would it cost to construct dark fiber to 95% of these sites, if it were done as part of a community-wide dark fiber build like that being planned in Singapore and already completed throughout greater Stockholm (and most of Sweden)?  Construction costs for point-to-point fiber are 20% or so above those of passive optical networks (PON), but the cost of the dark fiber alone (no electronics) is perhaps 80% of the total construction cost, so they're in the same ball park.  Verizon has directly and indirectly reported various costs for their FiOS construction (FiOS is a PON network), but all current estimates are below $1000 per home passed. Even at $5K per cellsite, the payback for a mobile operator would be measured in months.

Comments

[Craig Plunkett]

That's great feedback from the conference. In my experience with end users, and blogospherians, especially the web services kind, there's a total disconnect with operational details of the network edge. The provision of ubiquitous access is dismissed as a "commodity". There needs to be an educational effort to inform folks as to how the internet gets to their house or mobile device, just like how water and electricity are delivered, and sewage is taken away. I always thought this gap was prevalent among non-techies primarily, thanks for the heads-up.

[Rudolf van der Berg]

Interesting story. In my opinion countries will need FTTC/H all the way, because it is a platform to build national connectivity on top. This connectivity will be wireless for the first yard, hopping onto a wired network as quickly as it can and freeing the airwaves in the process. In this vision wireless networks like wimax, LTE, picocell, wifi are just plug in end points on top of a fixed fiber network.

Another observation to be made on this basis is. If fibre networks can be build for 30 euro/dollar per month. If there is no quality difference in the physical glass in consumer, business and Backhaul fibre. Than it would follow that a plain standard connection to the network would only need to cost €$30 per month regardless of whether you are a consumer, a corporate or a cell phone company. The only thing different is the type of access bought on top.

[Brough Turner]

Rudolf, Agreed! except perhaps for your final sentence.

Dark fiber lasts for decades, but the electronics that lights that fiber is functionally obsolete within a few years. That suggests there is a fundamental difference between the provision of dark fiber and the group that lights that fiber. I would like to own or control my own fiber so I get to select the ISP (from a competitive market) that I want to have light my fiber.

I don't want "the type of access bought on top" to come from the dark fiber provider.

[Rudolf van der Berg]

Ah, but then we are in agreement, because I didn't mean that it had to be bought from one source. I don't even mean that the electronics at both ends have to be from the same source. I quite like the Stokab model now that I'm getting used to thinking about it. (I used to be a managed up to layer 2 guy myself, because that is what we did in uni. There we worked for the good of all, but somehow that doesn't translate to the real world)

So yes please allow customers to buy the fibre from who they want. BTW nyquist capital expect a boom in fiber to cellsites based on what you write here. I think they are wrong where large parts of the US are concerned. If there is no adequate broadband service in an area, then there is no fiber in that area, therefore there will be no fiber to the cell site (as the mobiletelco is not going to pay for fiber all the way into Corn, OK.)

In the end life is simple: Without Broadband there is no Mobile Broadband.

[Rudolf van der Berg]

A quick reaction to expand my point on my blog here http://internetthought.blogspot.com/

[Brough Turner]

The full URL of Rudolf's post is http://internetthought.blogspot.com/2008/03/no-broadband-means-no-mobile-broadband.html

The other approach to backhaul, which I've seen in the US and several other countries where backhaul is only available from the monopolist, is point-to-point wireless backhaul. This can be unlicensed in the 5 GHz bands or licensed at 3.65 GHz, 24 GHz or even 60 or 80 GHz. The capital cost is not cheap, but usually much less than the cost of fiber (if fiber is even possible).