The Sky Is the Limit

The Internet has moved so fast that technologies just a couple of years old are made obsolete by new developments. The rush to bring killer ideas to the marketplace has engendered some truly competitive and impressive products. On the other side of the ledger are those ideas that are so truly goofy, so hard to believe, that you have to wonder how intelligent people even considered suggesting them. The wonderfully nutty Microsoft Bob belongs in this category.

On first blush, using satellites to connect to the Internet seems to belong to the same family of peculiar ideas. As if the Internet isn't slow enough, using satellites would introduce the delay, called latency, that is caused by waiting for packets to travel to the satellite and return to Earth stations. While much of the Internet's traffic travels toward user workstations, every mouse click on the Web and every outbound mail message requires that connectivity be preserved in both directions. Is every home and business going to become a satellite Earth station?

Still, in certain cases, I can imagine that satellites might make sense. Satellites already reach places that terrestrial lines never will: Portable Earth stations have reached Everest, one-person boats adrift on the Pacific Ocean and even remote expeditions on the Arctic ice pack. Each of these required highly specialized equipment and arrangements with specially outfitted Internet service providers. The expense of these connections is enormous, which leads me to wonder how satellites could ever be a part of a general-purpose plan for Internet connectivity.

Data communications using satellites is nothing new. Today you can purchase a Earth-to-satellite multicast link that delivers 24 Mbps and a point-to-point link that provides 1.5 Mbps. Unlike traditional terrestrial connections, a satellite user pays for a specific period of time on a specific satellite. That makes the technology good for very predictable communications requirements, but it seems to rule out typical Internet connections. For on-demand networking, a new approach was needed.

That new approach emerged from technology research at NASA. In 1993 NASA launched its Advanced Communications Technology Satellite (ACTS) that found a way to deliver hundreds of megabits per second. Not only did NASA prove that a satellite could be used for general-purpose networking, but it also let interested organizations give the service a try. At the beginning of this year the Federal Communications Commission granted licenses and spots in orbit to 13 well-financed and extremely interested companies.

Some of these companies are working to launch satellites into geosynchronous orbits so that they stay in lockstep over a particular geographical region. To accomplish this, the satellites have to be placed more than 22,000 miles above the earth, which leaves them, as the photon flies, an eighth of a second away from Earth stations. With a built-in latency of 0.25 seconds for every packet, any application that depends on real-time features or interactivity is probably doomed if it uses this kind of connection.

One solution is to simply build a network of satellites that roam the skies closer to Earth. That's what companies such as Kirkland, Wash.-based Teledesic and Motorola of Chandler, Ariz., have in mind. Using satellites that are less than 1,000 miles from the surface of the Earth, these companies plan to solve the latency problem by minimizing the distance between the satellites and the network endpoints. Unfortunately, because the satellites are closer, they cover a much smaller geographic region. As a result, a much larger number of satellites are needed to cover the same area. Worse yet, the satellites zoom across the sky instead of staying in the same place overhead. This requires a complex strategy for handing off signaling from the satellite serving you that is leaving your region overhead to the next satellite in the network.

In the case of Teledesic's network, more than 280 satellites will be needed. At a bargain basement $20 million per satellite, Teledesic's network will have nearly $6 billion is satellite costs alone. Even so, satellite networking vendors believe they will be able to offer impressive bandwidth at about the same price per bit as terrestrial lines. That may be true for terrestrial lines in Kenya or in the mountains in Western Montana, but it seems pretty unlikely for the already well-served areas of North America, Europe and Asia.

Despite this financial uncertainty, hundreds of billions of dollars are being invested in the competing satellite networks. Avoiding the costs of laying terrestrial lines is so attractive that one company, Angel Technologies of St. Louis is planning a wireless Internet access network that is just slightly more down to earth. Instead of satellites, Angel plans to have High Altitude, Long Operation (HALO) airplanes fly in a circular pattern high above metropolitan areas and relay packets between enterprises and their Internet service providers. Already satellite vendors are looking to form partnerships with Angel Technologies to take advantage of the new bandwidth available in areas of Internet traffic congestion.

These vendors plan to bring their networks into operation just after the end of the millennium. Will the enormous technical challenges of providing useful satellite connectivity be solved at a price that will make sense for general-purpose use? The satellite vendors are betting fantastic sums of money that the answer will be yes. I can't help but wonder that they've invested in the Internet equivalent of Microsoft Bob -- Mark McFadden is a consultant and is communications director for the Commercial Internet eXchange (Washington, D.C.). Contact him at mcfadden@cix.org or via www.cix.org.