## Too Many Networks?

Sunday, June 03, 2007For the past two years, I have been to New Zealand to speak at a great conference called Convergence, a Wireless and Broadband Forum event. My visits were hosted by the New Zealand of Trade and Enterprise and each year I had an opportunity to talk with the Communications Minister, David Cunliffe, as well as hear his speeches at the same event.

This year, Mr. Cunliffe spoke about auctioning 100 MHz of spectrum in the 2.3-GHz band and how that could help New Zealand reach its goal of providing broadband access to all of New Zealand’s four million people.

He told me the government would not dictate the technology to be used in this spectrum and while WiMAX was of interest to a number of companies that wanted to bid on it, it would not be limited to WiMAX.

On June 1, I received an email from one of my contacts in New Zealand with an article about the government’s revised plans for this spectrum. It has now decided to combine the auction of the 2.3-GHz band with spectrum in the 2.5-GHz band. The rationale for combining the spectrum auctions is to “offer six different companies sufficient spectrum to launch a national WiMAX service, while still reserving some for community use.”

This move goes back to a question many have been wrestling with: How many providers can be supported by a given population base? I guess we are going to find out. Today, there are already some pre-WiMAX systems (point-to-multipoint) in the Auckland area and Wellington has had a fiber network for more than ten years now. When I was there, the talk was of using wired, fiber and wireless to blanket the country with broadband.

New Zealand has no cable operators to speak of and most people receive their TV signals over the air or via satellite, there are two wide area-wireless operators (Vodafone and Telecom New Zealand) and a number of Wi-Fi providers (but no muni-Wi-Fi in anyone’s plans).

Some Basic Calculations

The numbers below are not representative of anything except a plain vanilla, back-of-the-envelope set of calculations. They are a gross average of numbers to make a point regarding the amount of revenue per cell site that could be reasonably expected. These numbers can be argued for months by many different people with many different views of the world—but for the purposes of this exercise, they make my point. If you disagree with my calculations, please feel free to post a comment.

New Zealand’s population of just over four million is spread out over two islands, although the bulk of the population is located on the North Island which is about 44,000 square miles while the South Island is 58,000 square miles. The combined total is about the size of ^{2}).

Looking at what is published about WiMAX coverage, which varies a lot, the most common guidelines I can find for 2.5-GHz WiMAX coverage in a rural environment is that it can provide 10 Mbps out to 10 km from the cell site and in dense urban areas this is reduced to 10 Mbps out to 2 km. Using standard math, the area of a circle with a 10-km radius is 3.14159 times 10^{2}, or 314.159 square kilometers. With a population density of 15 per square km, each site could provide coverage to 4,712 people (approximately). However, if we use the math normally associated with cell sites—that is the calculation of a hexagon (2.6 X Radius Squared), which removes the overlap between sites and is more accurate in predicting cell site coverage, the numbers work out to be 260 square kilometers per cell site. In this case, each cell site has the potential to provide service to 3,900 people.

If we use the average household size of three people, we have 1,300 households that could be covered from a single cell site. Remember that this is with a total data capability of 10 Mbps. If we convert the site into a 3-sector cell site (additional cost for antennas, cable, type of tower and other factors) we can deliver 10 Mbps in each sector and then only 434 homes would be contending for the 10 Mbps of data services.

These calculations are not expected to be on the mark for each cell site, only an average from which to work on the next issue: How much money (revenue) can a WiMAX provider expect to make on a per-cell site basis? If the attach rate for broadband services is 50% for the first three years, then the total market for WiMAX in a given cell site is 650 households. If each household spends $40 per month on broadband connectivity, the total monthly income per cell site is $26,000 NZ, not bad.

Now the problem is that this discussion was for a single WiMAX provider, with no competition from wired or other wireless sources. If we add five more WiMAX networks using the same model as above, and we assume each operator will get 1/6 of the total available market, the income per cell site goes down to $4,334.

But wait—there’s more! Since we have competition, as everyone knows, the price per household will be lower. If we deduct $10 per month per customer, the numbers now become: one network operator would get $19,500 per month while six networks would get $3,250 each per month per cell site.

On the cost side, you do the math:

Capex-

The cost of building out the site

The cost of building the back-end infrastructure

The cost of customer-provided equipment

Other

Opex-

The lease of the tower space

The cost of electricity

Insurance on the site

The cost to provide backhaul for the site

Other

There is also math available to figure out how many houses in each sector will be competing for the 10 Mbps of total data per sector, and math that will provide us with a pretty good idea of how much backhaul capacity will be needed for each site. I won’t go into that here, but no matter what technology we are talking about, I don’t see how six networks all offering basically the same types of data services can survive, let alone make money.

10 Mbps of shared data capacity for 524 homes (for one network or 87 homes for each of six operators) doesn’t seem like a lot of data capacity to me. Perhaps for Internet access, given the average usage, it might work, but if the vendors intend to try to sell video and other services that require a lot of data, I don’t think many of the customers are going to be very happy. The only way to get more capacity per home is to add more channels to each cell site or to build more cells closer together, both of which have financial implications.

For those who want to question my math, before you do, visit: http://www.wimaxforum.org/technology/downloads/WiMAXNLOSgeneral-versionaug04.pdf, open the PDF and go to page 9 where you will find the numbers for a system built on 3.5 GHz of spectrum, and look at the distances being shown by the WiMAX Forum. For indoor self-install customers, it is claiming a radius of from 0.2 to 0.5 km for a “standard” WiMAX cell and 1 to 2 km for a full-featured cell site. The numbers I have used at 2.5 GHz are generous and will, of course, vary from site to site, but the conclusion remains the same: Making money on these systems is going to be a difficult proposition, and enabling six vendors to compete on a nationwide basis is going to make it very improbable.

I am not picking on WiMAX here, I am picking on any data-only network with similar coverage and capacity constraints. How many networks can survive in a given environment will continue to be a question that eludes an answer for some time to come, but if I were an investor and I had been hyped on the wonders of this new technology, and the fact that broadband everywhere is a great goal, I think I might find somewhere else to invest my money.