When Alexander Graham Bell made his famous call to Watson, he probably wanted to discuss building the last mile of his planned telephone network. At least from that time forward, whenever two or more telephone engineers have gathered, they’ve talked about the last mile.

At first, they talked about how much copper to use relative to iron, which was much cheaper. Later, they discussed how to mix cable gauges and whether to use amplifiers, adapters, remote switching units, various digital devices and residential gateways. Most recently they have debated whether to use fiber, and how close to the customer to bring the fiber.
 

Enter New Services

In the past, the issue was always providing “plain old telephone service” -- POTS, in telephony lingo. Now the debate has changed in a very real way: It is about meeting requirements for new services.

The question is not just, “Which is cheaper?” Network providers are also asking themselves:

• What services will I provide?
• How much of each service will the customer need?
• How much bandwidth do I provide for these requirements? 
• How will compression advances affect my choices?

 
Adding these new questions makes the last-mile debate much more complex. To see how complex it is, one just needs to note that the three major telcos – Verizon, AT&T and BellSouth – have studied this issue with all of their great resources and come up with three completely different answers!

High-speed data access (mainly via DSL from telcos and cable modems from cable companies) started the change in last-mile forecasting. Dial-up access was just another form of voice, and the last-mile discussion was still about how to provide, not about what to provide. With the advent of high-speed access, the question started leaning toward ‘what.’ How much bandwidth was needed for the service to be considered high-speed? How much sharing could be allowed?

Video further tilted this discussion to ‘what.’ At first, this wasn’t apparent. Much of the early work – for example, test beds run by the telcos and cable companies – simply assumed that the answer to ‘what’ was television, specifically a single channel of standard-quality video viewed in the customer’s living room.

However, the telcos are now discovering that the real question is ‘what,’ and that the answer to that question will determine the ‘how.’
 

What Will Customers Do with Video?

Customers don’t demand bandwidth. Customers demand services! These include multiple channels of standard video; multiple channels of high-definition video; personal video recording (PVR or, as it is sometimes called, DVR, for digital video recording); video on demand; simultaneous use of multiple channels; and simultaneous use of high-speed data with video channels.

So to forecast bandwidth demand for video, we need to know how customers are going to use video. How many channels will there be? What kind of channels? How many will be on simultaneously? How much sharing can be permitted? What else must coexist with the video? How will the channels be used? The answers to these questions will determine what the customer wants from a video offering, and it is increasingly not just ‘television.’

It was so much simpler when all designers had to consider were the technologies available to deliver POTS, but those days are gone!

Since video is going to be the major driver of bandwidth, with high-speed access a distant second and voice so small as to be inconsequential, our forecasts must use as much information as is available about television and high-speed usage.

From a variety of sources, we find that:

• Americans view about five hours of television a day.
• About 98.5 percent of American homes have at least one television.
• About 48 percent of U.S. homes frequently have more than one TV on simultaneously.
• Prime-time viewing (8 p.m. to 11 p.m. Eastern time) includes about 75 million of our 110 million households, or 68 percent.
• Between 15 percent and 20 percent of U.S. homes have high-definition televisions (HDTV).
• Over 90 percent have VCRs.
• Between 13 percent and 15 percent have DVRs.
• The average home has more than three television sets.
• At least 50 percent of homes with both TV and high-speed Internet use both simultaneously. 
  

A Prime-Time Scenario for 2006

Taking this data, we can calculate that in a prime-time hour for 2006,

• 68 percent of households are watching one or more television set.
• 33 percent are watching two or more sets.
• 10 percent are watching HDTV.
• 10 percent are recording one channel on a DVR.
• 50 percent are using high-speed Internet access.
  

Figure 1: 2006 Bandwidth Requirements

Note: Voice is ignored because of its small impact.

The data in Figure 1 indicate that a BPON network (the type of network Verizon uses for its FiOS system) with 622 Mbps of capacity would still have plenty of capacity even if all 32 customers turned their television sets on during prime time.
 

Looking Ahead to 2010

Over the next few years, more people will be buying high-definition televisions, subscribing to high-speed Internet service and using DVRs. As standard television sets are replaced with high-definition sets, fewer people will be watching standard television channels.

What will these numbers look like in five years? We have to recognize that any approach to forecasting is full of uncertainty. It is, after all, a forecast of the future, and almost by definition it is going to be wrong! But forecasts can still be useful as a guide against which to measure progress and make corrections.

If we repeat this scenario building process several times using the assumptions outlined above, we can create some forecasts about the demand for bandwidth. Figure 2 shows the result.

Figure 2

Of course, the procedure is not quite so straightforward. We also have to consider (that is, forecast) the effects of better compression algorithms and the ability of some technologies to share link capacity among multiple customers.

Still, we know enough to conclude that, over the next five years, high-definition television and DVRs are likely to be by far the biggest drivers of bandwidth demand, high-speed Internet access will increase, standard television will fade away, and voice – once the telcos’ only service – will become an afterthought, not even visible on the chart. Planners’ main concerns today are not voice communications but rather trends in home entertainment, such as the speed of HDTV adoption and the spread of DVRs.

Mr. Watson, come here – I really need you now!

About the Author

Clifford R. Holliday is President, B & C Consulting Services, and author and analyst for Information Gatekeepers, Inc. Material in this article is taken from the author’s new report, “How Much Bandwidth Is Enough in the Access Network? – Strategies of AT&T, Verizon, and BellSouth in the Design of the Last Mile.”  The report is available from Information Gatekeepers Inc. at www.igigroup.com. Comments can be sent to the author at c.holliday@ieee.org.