|Sunday, 21 September 2008 23:39|
Can GPON be Effectively Unbundled?
EU Commissioner Reding and others suggest GPON will make it very difficult to share with competition, because all the signals go to a passive box in the field and there's no way to allocate a line to each ISP. GPON is probably cheaper than point-to-point Ethernet but the differences are less than 10% most places P-to-p can more easily be upgraded to higher speeds, hence "future-proof."
Quick answer: Traditional unbundling - a line from the exchange to each home for each ISP - does not work with BPON or GPON. Most experts think it's impossible, and some form of the mis-named "bitstream access" is the only practical method to maintain competition. Bitstream is generally considered a poor substitute, not appropriate long term.
However, "bitstream" might become interesting because GPON is close to completely congestion free up to a very high level. Rather than handing off the "bits" with a traffic charge and some limits on the customers, a "virtual line" can be created from the exchange to each customer with virtually unlimited bandwidth.
If the competitive carrier is charged per bit, as Deutsche Telekom requests, competiti prohibitively expensive for other operators to offer higher speeds/higher caps than DT itself. Very smart policy is required to avoid that trap.
Approach: Review existing deployments and expert opinion. Given those results are discouraging, look to further possibilities, not generally considered before declaring sharing GPON is impractical.
Results: Universally discouraging on sharing GPON. Neither BPON nor GPON have been effectively unbundled anywhere in the world, and the the near universal expert opinion is that it is impractical.
There is a proposal in Holland to do the splitting at the exchange in order to hand off an individual fiber for each customer, because the incumbent strongly prefers GPON. That would require an individual fiber to each home from the exchange raising the cost probably above active point to point fiber Ethernet. While they would be using GPON technology, the deployment costs will be similar to active Ethernet because of the need to run a fiber to every home. If the GPON sharing works effectively at the exchange, that would seem to make it fine from the regulator's point of view. This is unconfirmed talk at this point, not even close to field-proven technology, and I know no other country considering it. GPON has advantages even if you still need to run fiber to every home. It requires only a single laser for 10 to 32 homes; point to point networks require a laser for each customer, raising power requirements. Millions of lines of GPON are going to Verizon, Hanaro, and elsewhere, which should yield economies of scale.
Results from other similar systems: No BPON or GPON deployment has been effectively shared, obviously discouraging. The "Bitstream access" approaches have been disapponting, partly because the alternative opperator has few ways to offer a service significant'y better than the incumbent. In Europe, they have often been a way for an operator to gain volume quickly and then move to full unbundling.
Possible sources of error: Bitstream possibilities require further research.
Policy implications: The burden of adding these stations is modest. Since they will induce some people to choose cable, there is additional revenue that in many cases will offset the costs. A dozen subscribers at $500/year adds $6,000 to revenue, so a $5,000 investment will be quickly repaid in many cases. This would be one of the easiest ways to get hundreds of new voices in to many U.S. homes, a policy goal
Requiring a rapid pickup of these stations would impose high costs on those carriers not equipped with SDV. 2009 is the digital transition and the beginning of heavy DOCSIS 3.0 deployment, when imposing additional requirements will be particularly burdensome. Small carriers may have higher costs, because the they don't have the necessary equipment in place.
Possible policy choices:
Current arguments from the principals: Maybe a monopoly is not such a bad thing, they are whispering. The
Unspoken policy issues: Are the operators choosing GPON to eliminate competition, when Active Ethernet would be a viable choice based on cost and technology? Will GPON yield an effective monopoly (or duopoly with cable?) If so, is allowing monopoly possibly the best policy, because the increased profits might fund the buildout? (That argument was central to the U.S. eliminating unbundling.)