Providing wireless high-speed Internet service to Americans in rural areas that are not served by wired or fiber broadband is a lofty goal, but the wireless industry should not set itself up with unrealistic expectations.
In his recent State of the Union address and subsequent speeches and policy statements, President Barack Obama has made it clear that his administration will aggressively promote wireless broadband services. That’s not hard to understand; the wireless industry has been one of the few bright spots during the current economic downturn. Keeping it fed with new spectrum resources and political backing from administration policy is one way to help assure the industry will continue to grow and innovate. But while we bask in the glow from this presidential seal of approval, it is important that we not set ourselves up for problems caused by unrealistic expectations. In short, it’s time that we define just what services wireless broadband networks can and cannot deliver at practical cost.
One of the President’s wireless initiatives is to provide high-speed Internet access to Americans who live, work or attend schools in rural areas not served by wire- or fiberborne broadband services. The ultimate success of this approach, and the challenges we face in making it happen, will almost certainly depend upon how we define rural wireless broadband service.
To illustrate, let’s do a little exercise in rural network design. First, let’s be generous and assume that we have a whopping 200 MHz that we can use for the downlink channels. We will continue to be kind to ourselves by further assuming that the network will be engineered to provide ubiquitous rural coverage only to subscribers in fixed locations, equipped with user device antennas of modest gain and directionality, located on rooftops and aimed with at least some accuracy at their serving base stations.
With these constraints, in order to manage mutual interference so that channel quality will support high level modulation, it’s (optimistically) possible that at any given time about one-half of the available bandwidth will be usable by any given base station, and that the downlink will deliver an average throughput of 3 bits per second per Hertz. Each base station will then have a seemingly gargantuan downlink capacity of 300 Mbps.
Our rural broadband network looks great from the capacity side. Now let’s consider demand. In wireless networks where users get essentially unlimited service for a fixed monthly fee (which could be the case for rural broadband networks), operational economics hinge on two basic factors: average per-base station costs for deployment and ongoing operation; and the ratio of users per base station. If the average monthly cost per base station (including amortization of capex) is $20,000, and the monthly user fee is $25, then the ratio of users to base stations must be at least 800 or the operator will lose money. Of course, we want our network to make at least a little profit, so let’s assume that each of our base stations needs to serve 1,000 users. That will work so long as the average per-user throughput demand during times of peak loading does not exceed 300 kbps. The “burst” speeds observed by individual users will be vastly higher, but on average, over the course of perhaps a minute or so, each can only consume an average of 300 kbps.
That’s more than adequate to support applications like Web browsing, email and downloading of individual photos, songs and eBooks. But video is another matter. Throughput requirements for streaming video may range from around 300 kbps for low resolution YouTube clips to perhaps 2 Mbps for conventional quality television to as much as 4-6 Mbps for high definition TV. Let’s say that, on average, each streaming video download requires 2 Mbps. Our sterling 300 Mbps base station downlink capacity will then support only 150 simultaneous streaming video users, which represents 15 percent of all users.
Will that be enough? Based upon the growing popularity of streaming video applications, probably not. Remember, we are designing a network to serve rural areas where there is no wireline broadband alternative. If our users are going to get streaming video on demand, it’s going to be delivered on our wireless network. And to do it in a way that’s economically sustainable, we’re going to need lots and lots of spectrum.
That brings us to another consideration in our design exercise. If a base station is going to serve 1,000 users, it has to cover at least that many people. More realistically, we can probably assume that the number of individual fixed broadband data subscribers will be about one-third of the covered population, so each base station will, on average, need to cover about 3,000 pops. Rural population density (not counting people in towns where, presumably, wireline broadband service is available) may be around 2 per km2. That means that each of our base stations will need to provide coverage to 1500 km2.
More realistically, we will need to engineer sufficient overlap between cells to assure reliable ubiquitous coverage, so to be on the safe side, let’s figure 2500 km2, which corresponds to a coverage radius of about 28 km. That will work fine for the plains of Kansas, but to get range like this in the hills of Appalachia, we will be restricted to using lower frequency bands. In general, it will be tough to make rural broadband economically viable if we are counting on using bands like 2.5 GHz, and what we really want is more in the 500 – 900 MHz range. So if we’re going to support allyou-can-eat video on demand and other datahog applications in our rural broadband network, we will need bountiful spectrum from a frequency band that doesn’t have all that much to give, regardless of FCC policies.
Of course, urban and suburban network operators are facing the same problem with data-hog users. But they have two advantages: a much broader range of practical spectrum and, more importantly, users with alternative (wireline) means of obtaining data-hog services. Urban users may grumble when usage caps or other, less clumsy measures make it impractical to use their smartphones for watching movies, but at least they continue to be able to do so in their homes, schools and offices.
The viability of providing ubiquitous broadband service in rural areas may hinge not so much on government spectrum policies as on how that broadband service is defined. If the model is that of wireless services in urban areas, where carriers will ultimately need to restrict video applications, the numbers look good. But if these networks are expected to provide service comparable to what users can get where wireline broadband is available, then things will be vastly more difficult.
Drucker is president of Drucker Associates. He may be contacted at firstname.lastname@example.org.