Bandwidth and Throughput Math

Part of my current research agenda involves a comparative assessment of U.S. broadband market penetration and next generation network deployment. I am glad to note that the incumbent wireline carriers, such as Verizon, belatedly have invested billions of dollars. But on the other hand, because I live in the hinterlands I can get broadband at staggeringly high rates compared to what other consumers play in U.S. cities and in other nations. So I see a mixed bag corroborated by the relatively poor comparative performance of the U.S. in unbiased measurements of market penetration, digital opportunity and network readiness conducted by the International Telecommunication Union, see http://www.itu.int/osg/spu/newslog/CategoryView,category,Broadband.aspx; the Organization for Economic Cooperation and Development, see http://www.oecd.org/document/9/0,2340,en_2649_34223_37529673_1_1_1_1,00.html#Data2005; and http://www.websiteoptimization.com/bw/0510/; and the World Economic Forum, discussed in a previous blog entry.

Part of the challenge in this research lies in normalizing the data, i.e., comparing apples to apples, and deciding what constitutes success. The data compilations consider most helpful identify the number of subscribers per 100 residents, a broadband measure of “teledensity,” and the cost per 100 kilobits per month. See the ITU’s Digital Life Internet Report, http://www.itu.int/osg/spu/publications/digitalife/.

But even before one looks at the data, it makes sense to achieve consensus on the definition of terms such as bandwidth and throughput. Ironically the much maligned Senator Stevens from Alaska got it right: the Internet is a bunch of tubes or pipes. For the Digital Literacy course I teach at Penn State I analogize bandwidth as the size of the pipe, e.g., half-inch bathroom pipe versus 12 inch water main pipes. For throughput the pipe analogy considers the number of gallons that flow through the pipes per minute.

I am trying to consider the value proposition of broadband access in terms of available bandwidth (dedicated or shared), the likely throughput and aggregate usage per month. This is where the math and the value proposition get curious. You might not know that cable systems typically allocate only 6 MegaHertz of bandwidth for shared Ethernet-type access to the Internet. So if I am sharing 6MHz or less with hundreds of other subscribers, my virtual bandwidth allocation is quite small, even though I get multi-Megabit per second throughput and “All You Can Eat” uncapped access. Bear in mind that dial-up Internet access derives about 50,000 bits per second (50kbps) from an allocated bandwidth of 3-4 kiloHertz. For DSL, the dedicated copper local loop bandwidth expands by about 1.25 MHz. See http://electronics.howstuffworks.com/dsl2.htm.

Internet access subscribers actually care little about bandwidth except for its impact on about throughput, the number of downloadable and uploadable bits per second. But getting back to the value proposition, the ITU (in the Digital Life publication) ranked the U.S. 10th globally in terms of prices per 100 kilobits per second for 2006 using a $20.00 monthly subscription. Because I pay double the imputed rate, my value proposition would rank about 18th globally. Of course if you can get multi-megabit per second service for less than $20 a month, your value proposition is better than the national average

So the U.S. has some ways to go before government and carrier officials can self-congratulate. Still no one in the U.S. can come close to the global best practices of $0.07 per 100 kbits/s in Japan. Put another way the average U.S. broadband price is 700% higher than the Japanese average.