Monday, December 10, 2012


Chattanooga update

Looking for the savings to appear in EPB's accounts the costs for year ending 30th June 2011 and 2012 show costs :- Other operation expenses - 37.4, 40.8m Maintenance - 28.2, 23.4m At least the maintenance is going the right way from one year to the next, although the previous 3 years were Maintenance :- 13.6, 15.6, 18.0 (2008-2010) so as yet we haven't seen a massive saving from "smart grid" operation have any beneficial impact on maintenance costs, if anything it's gone the other way by 33%. Other operation expenses have risen above previous levels too :- Other operation expenses :- 38.0, 38.9, 37.5 (2008-2010) Moving onto the fiber optic operation, the residential revenue has grown healthily from $42 to $61m pa for a $10m pa increase in operating costs.

Thursday, September 08, 2011


BT Tweak sub-loop unbundling prices

Earlier in the year OFCOM ruled that BT needed to remove some cost elements from some of Openreach's charges for sub loop unbundling. Openreach have now notified OFCOM of the revised charges, which result in one price decrease and one increase with effect from 01/12/11.

The saving arises in the Shared MPF sub-loop product, where the price to connect a shared sub-loop to a CP's LLU equipment will fall from £127.61 to £115. This is the cost of hooking up the broadband part of a subscribers service to a local LLU cabinet while leaving the voice part connected to BT's telephone exchange, covering both transfer of existing lines and provision of new ones.

The price increase is in the provision of a new local subloop for the fully unbundled MPF product which increases to £127.61 "so that it properly reflects the costs of provision and thereby aligning it with the price for SLU-MPF transfer (which remains unchanged)".

No changes to the annual rental for shared and fully unbundled paths are proposed, so they remain at £11.47 and £93.96 per annum ex VAT respectively.

I was under the impression that OFCOM identified cost elements in the fully unbundled loop pricing that related to the E side from exchange to cabinet and argued that these should not be included in the rental of a fully unbundled sub-loop, but this hasn't passed through as a reduction in rental or connection charge.

None of the changes will satisfy current or potential sub-loop unbundlers, they are still left looking at a competing Openreach FTTC product where the connection charges are £80 for all flavours of FTTC and FTTP ! How can a fully costed shared sub loop connection charge for LLU come out much higher than connecting the same sub loop to an Openreach DSLAM and providing that DSLAM port and backhaul ?

When it comes to disconnection, you can cease an Openreach FTTC circuit for £5.37 but ceasing a shared sub-loop costs £100.67 !

Labels: , , , , , ,

Monday, August 01, 2011


Chattanooga FTTH economics

I previously noted the rather high cost per premise of EPB's Chattanooga FTTH deployment. While this scheme is the poster child of the UK's "FTTH or bust" lobby it does look expensive in terms of cost of installation and the retail pricing would be a bit rich for our tastes too.

Obviously it's a sexy proposition to have a gigabit connection, even if only 12 residents take that service. One has to question the large debt involved in the current climate - a company with a turnover of $480m in 2008 taking on a fibre-optic bond debt of $220m doesn't look like something that would fly in the private sector. $220m divide by 169,000 residents is a debt of $1300 per head which if transferred to the UK population would be $78 billion or £47.8 billion which is 50% higher than the Analysys Mason estimate of under £30bn for point to point FTTH throughout the UK.

EPB's premise was of reducing the cost of operation of it's electric grid. The accounts show the following costs for 2008, 2009 and 2010 in $m :-

Other operation expenses :- 38.0, 38.9, 37.5
Maintenance :- 13.6, 15.6, 18.0

So it's early days but the tens of $millions of smart grid savings aren't showing up just yet.

The operating expenses and revenues of the fibre optic system are starting to take off :-

Fibre optic operating expenses :- 9.8, 11.0, 16.1
Fibre optics sales :- 15.5, 17.2, 20.9

It will be interesting to see the full picture as the system rolls out and full year accounts reflect a mature customer base. Adding $5.1m in operating expenses to gain $3.7m in revenues from 2009 to 2010 doesn't look great but the accounts of large companies don't provide a whole lot of detail.

In 2010 only 18% ($4m) of the fiber optic revenue was from residential services. When 30,000 customers paying $100 a month are fully reflected in the accounts we should see perhaps $36m of revenue making the total more like $58m with 62% from residential sales.

I look forward to the 2011 accounts, and would be happy to be corrected if I have misunderstood any of the above. It's a shame proponents of schemes like Chattanooga don't look beyond the willy waving gigabit speed into the detail in order to see if the thing is remotely affordable or sustainable.

Labels: , , , , ,


Electric Broadband

The twitterverse has carried links to a recent audio interview with James Ingraham an employee of EPB, a US local power utility based in Chattanooga, Tennessee, describing their diversification into communications and specifically fibre optic connections to residential properties FTTH.

Things are different over there, with lots of small municipality utility companies when compared to the UK's regional distribution network operators. EPB talks of being one of the biggest in the US, serving 169,000 residents in a 600 square mile area - that's a square 24.5 miles on each side, or a circle with a radius of 13.8 miles, so tiny compared to the UK operators. EPB is also what we would call a quango, or perhaps a council owned non-profit business, so it's part of the public sector unlike the UK's wholly privatised electricity industry.

I have extracted some information from the hour long interview, and added some internet research to fill the gaps, with a view to helping understand what EPB are doing and it's relevance (or otherwise) to the UK situation.

EPB have spent about $300m on installing a gigabit fibre optic network to connect up their electricity distribution assets to form a "Smart Grid" and at the same time to provide residential triple play services and business services. This project has been underway since 2008 and gigabit residential services were announced in September 2010 - so far there are "a dozen" customers taking this $350/month service.

Much of the investment was based on the Smart Grid delivering savings on meter reading, reducing stolen electricity, faster outage recovery, better design etc. A Federal grant of $111 million was provided for this aspect ($656 per resident) and a bond issue completed in April 2008 raised $220m. Personally I have no idea why a SCADA system on a power distribution network would require gigabit connectivity rather than something more like RS422 but we'll leave that to one side. Suffice to say that a belief in large savings ( $30m pa) on the electrical side drove at least $160m of this investment.

EPB's broadband products don't come cheap - 30 Mbits/s standalone symmetrical broadband at $58/month "plus taxes and fees" is probably very good value but the equivalent price point in the UK would be £58/month inc VAT and to be honest that's going to be a hard sell, as Virgin Media have found with their 50M DOCSIS 3 cable product. My £1 = $1 conversion reflects the reality of retail pricing either side of the Atlantic - if you look at the Federal minimum wage, EPB's power price per kWh or the price of a Big Mac it's invariably $1 of US street price equates to £1 UK street price.

The high pricing also has to be seen in the context of limited competition in the US broadband market - Comcast charge $100-$117/month for 50M downstream / 10M upstream cable broadband. EPB have introduced a third broadband supplier to compete with Comcast cable and AT&T DSL. Another "electric broadband" operator in Tennessee also knows how to charge - $160 per month for 16M down / 2M up service.

EPB fibre has 30,000 residential and 1,300 business customers. It isn't clear what proportion of the customers are covered but Ingraham referred to building "where population is densest and working out". About 50% of electricity meters are connected by fibre, they only run fibre where the residential customer takes broadband, TV, or phone service and the rest of the electricity meters use 900 MHz mesh networking to communicate to those with fibre connections.

169,000 residents at 2.5 people per home would give 67,600 homes and 50% of that is 33,800 so it looks like about half of the homes take the fibre service which fits with the 50% metering comment . $300m is $4,438 per home passed or $8,875 per home connected which is a massive sum compared to other FTTH projects like Verizon's FIOS. Obviously the accounting of a dual purpose project gets complicated - how much of the spend was necessary to deliver the electrical savings, and how much to deliver the fibre optic communications services ?

Sunday, July 17, 2011


Community Networks or Council Broadband ?

Recently I've been looking at US implementations of "community networks" after our local "fibre or bust" proponents were tweeting links to them.

Some of them didn't look like community networks, at least not by my personal understanding of the term. Perhaps we need a more granular taxonomy than defining everything that isn't owned by a large corporate as "community".

An article in the Huffington Post from an advocate of community networks puts forward a campaigning stance that "community networks" are at a disadvantage compared to the one or two large corporate providers present in a typical US locality (cable provider and DSL from incumbent telco). The network in question is a municipal network "Salisbury Fibrant"which offers triple play services over Fibre to the Home in Salisbury, North Carolina in competition with 7, 10 or 15M Time Warner Cable and 6M DSL from AT&T.

I call this a "municipal network" because it is financed, owned and operated as part of the local government, with some controversy as one would expect in the freedom loving and capitalist USA. In essence it's offering a 15M/15M symmetrical FTTH service at $45/month with $20/month increments to 25, 50, 75 or 100M so the 100M service is $125 per month taken on its own. There are a wide array of TV and phone bundles too. The DSL and Cable competition do look poor and appear to have dropped prices in response to Fibrant's arrival.

Those opposed to the Salisbury Fibrant project point to low takeup, high retail prices, mounting debts and increased taxes and local utility fees to pay for Fibrant. It's hard to judge the truth of the matter from this distance, but time will tell.

In nearby Wilson, NC the Greenlight municipal FTTH network has disconnected 1,000 customers who never paid a bill - an alarming figure relative to their 5,000 or so subscribers.
In the State of Vermont the city owned Burlington Telecom got into financial problems when the takeup rate for its fiber system didn't meet requirements and losses mounted. So all is not well in the world of council broadband.

In case anyone runs away with the idea that things would be different in the UK, take a look at Swindon's controversial municipal Wi-Fi project "Get Signal" provided by Digital City.

I'm sure there are proper community projects in the USA that are not council broadband, and look forward to reading about them. All links and hints gratefully received.

Labels: , , , ,

Friday, July 15, 2011


2M USC and Rural Digital Divide

A draft Requirements Document put out by BDUK contains a couple of interesting snippets, at least in its current form. The invitation to comment has now closed.

The 2 Mbits/s Universal Service Commitment is now being expressed as "a minimum edge-of-network throughput of 2Mb/s and that most solutions will exceed 2Mb/s". Throughput is defined in the document as being measured by transferring a file and measuring the time taken, so it is the actual useful data throughput without overheads rather than the sync speed, IP profile or other higher level measure.

Interestingly, this means that a 2M fixed speed ADSL service will not meet the USC requirement as expressed, because the measurable throughput of IPStream Home 2000 is at best 1900 kbits/s. Current MaxDSL services will need an IP profile of 2500 to meet the spec which is a sync speed of at least 2848 kbits/s at the ATM level.

In the same document (p19) an example of a Local Authority call-off shows the speed vs rurality hierarchy at the core of BDUK's plans :-

5A1 Delivery of 50Mb/s to xx Market Towns (Example) L2
5A2 Required for delivery of 20Mb/s to 50Mb/s to xx Villages L2 (Example)
5A3 Delivery of < 20Mb/s to xx Hamlets (Example) L2
5A4 Required for delivery of xx Community Hubs (Example) L2
5A5 Required for delivery of 2Mb/s to edge of network access speeds L2
to xx Rural Communities (Example)

So if you live in a Hamlet (OS definition " small, isolated group of houses without a church. ") then it's ADSL2+ or less for you. Granted it's an example, and the document is a Draft, but it does suggest what we have to look forward to.

Another recent BDUK publication is the Data Model Explanatory Notes which accompany a model that appears not to have been published. This model looks quite interesting as it calculates funding requirements and costs for deploying various broadband solution. It also reinforces the notion described above that isolated and sparse rural types will be getting BDUK's slowest offering :-

" The superfast broadband model therefore calculates costs and revenues for the final third on a cabinet by cabinet basis. The cheapest cabinets in terms of investment gap per customer served are upgraded first until a floor threshold for fibre coverage (e.g. 90% of premises) has been reached in every Local Authority area and fibre connectivity to every community has been provided.

Premises that are expected to receive less than 2Mbit/s, after targets for fibre coverage have been reached, are input into a wireless and satellite cost model. For 10x10km grid squares with a population density above a chosen threshold it is assumed that a wireless mast is constructed, and all premises in grid squares below this threshold are served with satellite. "

From this I conclude that if you can currently only get 2M ADSL then you'll be offered a wireless solution if the population density is sufficient, or a satellite dish if not.

Labels: , , , , ,

Tuesday, June 28, 2011


Digging further into "Propel Northants"

Trying to untangle the loss making assumptions behind the DCF analysis of the "Propel Northants" project I have observed the following questionable assumptions :-

  1. Take-up aspiration is very low - "a total connection rate of 30% of the total private sector premises passed in year 2026"
  2. Rate of take-up growth is abysmal - "2% of all existing Private Sector premises (B Type and non B type) passed are connected each year"
  3. A wholesale only "Open Access" model is used, with a modest proportion of revenue captured as a result - "it is assumed that the wholesale network operator takes 50% of the total retail revenues achievable"
  4. Retail pricing model is unambitious, matching first generation broadband pricing with a SOHO office internet / telephony bundle at £35 per month
  5. Large parts of potential market are excluded - "Other Public Sector premises (i.e. Education and Health) do not connect to the network and procure their telecom services via an alternative route"
The network is a substantial investment and isn't going to be achieved by community activists digging trenches - 110 km in 10 distribution rings are proposed, connecting to 224km of core fibre ring with 8 points of presence.

Capital costing is in line with corporate civil engineering costs, as one would expect from a report prepared by Atkins and The University of Northampton. The figure of £1,309 per business and public sector premise passed is perhaps not unreasonable, but escalates to an eye-watering £7,644 per premise connected when the low takeup assumptions are factored in.

"Propel Northants" could represent a good launch platform for further growth of FttH services in Northamptonshire, but its current economics leave it stillborn.

Can the above weaknesses be realistically addressed to make the project an attractive investment ?

Labels: , , , ,

Thursday, June 09, 2011


How to lose money fast ? public sector fibre optic broadband

I'm reading a report prepared by Atkins and the University of Northampton which considered the costs and benefits of building a fibre optic broadband infrastructure in Northamptonshire to produce economic growth - Propel Northants.

As a numbers man I have been digesting the discounted cash flow (DCF) analysis which shows a lot of red ink in Table 33. The project starts with £30m of capital expenditure in Year 0 and then makes a loss every year for 15 years, producing a NPV of (£57m). This says that the project is the same as burning getting on for £60m in a bin outside the council offices.

To be fair the annual loss does decline through the project, and is only a million quid by 2026, but it is still a loss from start to finish.

I will write more on why I believe the DCF produces such a dire result as I get my head around its assumptions. One obvious factor is that Health and Education are assumed to do their own thing and are therefore not included as potential revenue customers of the fibre network. This is a shame as a) these are two of the biggest players in the public sector and b) the model assumes that the fibre network gets 80% of the retail revenue from public sector customers (as opposed to 50% from private sector) with its open access wholesale model.

This page is powered by Blogger. Isn't yours?