If you didn't see the first wave of affordable, high-speed Internet connectivity coming, you have nothing to worry about. Heck, the DSL extravaganza, at least at the consumer level, is three years old, and the market appears to have more losers than winners (see The Worst Show on Earth). It will take several more months—or years—for the DSL market to shake itself out. In the meantime, selling DSL service is rapidly becoming a commodity. Is there a way to capitalize on what's coming down the pike? You bet. The next wave of transport technology will bring the Internet nearly everywhere on the planet at breakneck speed.
We're talking wireless here. Specifically, fixed wireless and satellite technology solutions that will enable you to deliver new service and connectivity options to your customers as well as partnership opportunities all over the place. So come along as we show you how to cut the cord.
Predicting the Future
Getting a handle on fixed wireless options takes some work. Technology-wise, there are the requisite new acronyms that you'll have to learn about, along with a little bit of physics. Politics factor in to an extent, because the FCC has a major say on who gets to pump what into and through the air. Economics also will play a big factor. You probably would have a problem bidding for your own personal frequency—the last auction in 1996 raised roughly $4 billion for the U.S. government from prominent players like AT&T, Sprint and WorldCom . Finally, there are several different approaches to delivering the Internet without wires, and each involves the use of a different piece of the spectrum. In some cases, a combination of approaches is used to achieve the objective.
Certainly, market analysts agree that the potential is there. According to a recent study published by Pioneer Consulting, the fixed wireless broadband market is expected to grow from $200 million in 2000 to $4.1 billion in 2003.
Let's get the acronyms out of the way. While there are myriad fixed wireless solutions in use, they all fall into one of two categories: those that operate below 20GHz and those that operate in the 20GHz to 40GHz range. Note that much of the spectrum is reserved for other purposes. A PDF version of the spectrum allocation can be found at www.ntia.doc.gov/osmhome/allochrt.html.
There's a lot of action taking place in the lower end of the spectrum. That's due to several factors, notably cost, range and licensing. Solutions operating at lower frequencies are relatively inexpensive to implement from components that are in widespread use, have a transmission range of miles, and—in most cases—do not require a license from the FCC. In this arena, there's a lot of attention being focused on a solution called MMDS.
MMDS refers to multichannel multipoint distribution services, a broadband wireless access solution that operates in the 2.5GHz band. Originating as a solution for video broadcast in the 1980s, the technology never caught on commercially. In 1998, the FCC authorized two-way transmission in each of the 6MHz channels (there are 33 such channels). Pilot deployments were conducted last year, and 2001 is when a lot of actual deployment activity will occur, according to Susan Heilman of equipment vendor ADC.
One of the big factors inhibiting mass-market acceptance is the cost of the customer-premise equipment, or CPE. Currently, MMDS-based CPE is well below $1,000, and Heilman sees it coming down in price to the sub-$500 range within a year. That begins to approach xDSL gear in terms of cost. (Some companies, however, like Cablevision of Long Island, are giving xDSL gear to consumers through special retail offers.)
In addition to cost, MMDS faces other challenges. For one, the FCC doesn't permit a self-install of this type of equipment, making it tough to aim for a consumer market. And finding installation experts with RF and IP skills is a challenge, according to Heilman. Still, the operating range is on the order of 20 to 25 miles with a 2Mbps to 3Mbps uplink and a 10Mbps downlink.
Analysts are bullish about MMDS's promise. Cahners InStat foresees a compound annual growth rate for a particular segment of the wireless market of 218 percent from 1999 through 2004. Not too shabby.
MMDS is not the only low-end game in town, however. Startup service provider Kite Networks, for one, is beta testing its own fixed wireless service in Phoenix. Kite's solution revolves around gear operating at the 5GHz, or Uniband, spectrum. The company has ambitious deployment plans. Solutions in Dallas/Fort Worth, Atlanta and Houston will be rolled out by the end of 2001. Assuming those deployments go well, Kite will push ahead with a "50-market plan," according to company VP Scott Tenney. Kite plans to offer a variety of connection speeds to customers, ranging from 500Kbps to 2Mbps. Each cell of coverage in the Kite network has a three-mile radius, and the company plans to use system integrators and resellers as agents for the service, as new locales are brought online.
Meanwhile, Metricom is pushing ahead with its Ricochet Network. Metricom's MicroCellular Data Network approach uses two Industrial Scientific Medical (ISM) bands of regulated, unlicensed spectrum: the 900MHz band and the 2.4GHz band, in addition to the licensed 2.3GHz Wireless Communications Systems (WCS) spectrum. Granted, Ricochet is designed for mobile computing, but the comparison with Kite highlights major innovation going on at the low end of the spectrum to deliver the Internet everywhere.
"The technology is evolving quickly," says Tenney. For example, just last week, MMDS equipment vendor Speedcom announced a technology licensing agreement with SRI International for PacketHop. The deal involves wireless routing software that overcomes the necessity for line-of-sight placement of transmitter and receiver. "The only barrier to wireless becoming the dominant last-mile solution has been line-of-sight," says Bruce Sanguinetti, president of Speedcom. "PacketHop solves the problem."
The action's not just at the low end of the spectrum, either. Local multipoint distribution services, or LMDS, refer to the access solution that operates in the 20GHz to 40GHz range. While that varies by country, the spectrum allocated to LMDS in the United States is in the 28GHz to 31GHz range. There's considerable financial interest in this band, as well: In just 26 days during 1998, the FCC raised nearly $580 million by auctioning the spectrum segment in 493 areas in the United States.
While LMDS operates at a much higher frequency than MMDS, solutions based upon this technology are far more susceptible to weather. Rain, in particular, can impede the signal. There are also distance limits; LMDS transmission range is on the order of one to two miles. The higher frequency means more bandwidth, making LMDS ideally suited to deployment in urban areas. LMDS-based systems are more expensive than MMDS systems, but when compared with digging holes and installing fiber in urban areas, they come up as inexpensive alternatives to delivering bandwidth.
The economics can be compelling. Michael Sanderson of Ensemble Communications estimates that putting a fiber loop in San Francisco could cost more than $10 million. By contrast, an LMDS base station capable of 1.5Gbps would cost on the order of $1.5 million. These calculations don't take into account the time it takes to get the connection deployed—a few days for the LMDS system, and potentially weeks to months for the fiber, unless the building's already hooked up to a metro fiber ring.
While LMDS, MMDS and the other land-based broadband wireless technologies are hotbeds of activity right now, they don't cover everywhere. To get complete geographic coverage, you've got to go into space—the realm of satellites—where there's a raging battle evolving. There are myriad factors to consider, just from a technology perspective. For example, should you use or recommend a system based upon a low earth orbit (LEO) model? These systems are good choices for highly interactive applications because they have little latency (the time it takes for a signal to travel from the earth to the satellite and back). However, because each satellite covers just a spot on the earth at any given instant, the LEO model requires many satellites to provide necessary coverage. Will it cover the geographic area that you need? Both SkyBridge and Teledesic think it will.
On the other hand, a geostationary earth orbit (GEO) system such as Loral's CyberStar provides near global coverage all the time, but has much higher latency. So it may not be the best way to deliver a chat application to someone. You'll need to match your application to the system that best supports it.
Another factor is bandwidth—again. Do you go with a system running the existing Ku-Band (10GHz to 18GHz) technology or wait for the next-generation Ka-Band (28GHz to 31GHz) devices to hit the sky?
Economically, building a space communications infrastructure is expensive. SkyBridge claims that it will spend $4.2 billion to get the space component running, and another $1.9 billion on the terrestrial infrastructure. Infrastructure providers are taking huge gambles that satellite-based Internet delivery will be economically viable. Make sure you do your due diligence concerning resources and capitalization when selecting a partner.
Will your business model support the cost of a satellite connection? SkyBridge estimates that base terminals at a user location will cost in the neighborhood of $700.
It gets confusing quickly, and the problem is that many of the systems described here won't be online for a few years. Teledesic, for example, indicates that service will begin in 2005.
There are some space-based communication services active right now—such as Tachyon (which we're in the process of evaluating) and SpaceData. Both of these services utilize an existing satellite infrastructure. Perhaps that is the safest way to proceed in this new environment.
Whatever you decide is the best way to go, via earth or space, you won't need a crystal ball to see that there is ample opportunity for participation at a variety of levels. That's because the pipe isn't owned by just one entity. There are a number of providers of both infrastructure and service that will be willing to help you out. It's the perfect partner environment.
Do you need to move large—and we mean large—chunks of data from a mobile location periodically? Three-year-old SpaceData International LLC may have a solution for you. SpaceData uses NASA's Tracking and Data Relay Satellite System (TDRSS) to haul data from nearly any point on the globe through New Mexico and Texas and then out to the Internet. The privately held company is leveraging the privatization of NASA's space-based and ground-based communications assets, which the agency was directed to commercialize in 1996.
SpaceData's first service is called SeismicStar. Aimed at the oil exploration market, SeismicStar collects and transmits terabyte-sized files of marine geophysical seismic data from acquisition vessels in the ocean directly to land-based supercomputers. The key to the whole operation is speed. With a transmission speed of 311Mbps, Jay Gnowles of SpaceData says, "We can transfer data faster than any other satellite system in operation."
With that kind of bandwidth, the company can transmit a 170GB file from the ship to the ground in a little under 75 minutes. Sounds fast—but that's only the beginning. With the deployment of newer technology under the guise of the TDRSS-H, -I and -J satellites, channels with higher speed (the Ka-band, operating at 23GHz and 25GHz to 27GHz) will become available. TDRSS-H was launched last June, while the I and J versions are projected to be launched in 2002 and 2003, respectively. SpaceData is preparing to test a 466Mbps transmission rate, with a 622Mbps solution already on the drawing board.
While the speeds are mind-boggling, the solution is not for everyone, due to the nature of the technology. This is not an always-on solution; rather, it's suited for an application where a large amount of data is collected and then periodically sent to another location. For a data transfer to take place, the TDRSS is aimed at the transmitting station, communication is established, and the transfer begins. When complete, the satellite then is aimed at a different location for another job. Call it "broadband-on-command."
The SeismicStar base station itself is a high-end setup. It includes dual Sun E450 servers configured with a StorageTek 9145 RAID array. Typical storage capacity is about 1 terabyte. In addition, there are dual ATM switches from Fore Systems and dual 155Mbps New Tec modems. A 2.4M Sea Tel marine tracking antenna enclosed in a radome completes the setup.
We're used to thinking of gigabit Ethernet as the technology of choice for linking together Fast Ethernet switches, locking together storage area networks and binding servers into clusters. It can be more. Much more.
Companies like Extreme Networks and Foundry Networks are taking Gigabit Ethernet from the back office to the wide area network (WAN). That's because Gigabit Ethernet is cheaper to run and implement than asynchronous transfer mode (ATM) or dense wavelength division multiplexing (DWDM). Analysts from The Yankee Group estimate that Ethernet-based Internet can be delivered at one-fifth the cost of other broadband Internet services.
There are other advantages. For example, with Gigabit Ethernet, you can use fiber to deliver variable Internet bandwidth to your customers. The Extreme Networks' Alpine "i" series switches can, for instance, deliver 500Kbps to 1Gbps traffic without any physical provisioning. Pretty impressive stuff.
In the future, you also can expect to see first-mile Gigabit Ethernet delivered over copper. The copper Gigabit Ethernet technology, IEEE 802.3ab standard (a.k.a 1000Base-T), has found a home in back offices, but its practical range is only 100 meters over ordinary Category (Cat) 5 cabling.
Extreme is pushing that range. The company now claims it can deliver Ethernet over T-1 and DS-3 circuits or even with very high-bit-rate DSL (VDSL) over Cat 5 at distances of up to several miles.
For now, though, optical fiber, usually in the form of a synchronous optical network (Sonet), will be the physical medium for most Internet deployments.
There are two broad ways to handle Gigabit Ethernet on Sonet. The first, Extreme's approach, is simply to use Sonet's fiber purely as a physical media and run Gigabit Ethernet over it, just as if it were a fiber-optic cable. Foundry, however, wraps Gigabit Ethernet frames within Sonet frames with its Global Ethernet approach.
You always end up paying one way or the other. In the case of Gigabit Ethernet, the cost is in quality of service and bandwidth control. Even the gigabit supporters admit that they lag behind ATM in these areas. But, the Gigabit Ethernet players are moving forward quickly on an Internet Engineering Task Force (IETF) standard called Common Open Policy Service protocol for Provisioning (COPS-PR), which will address those shortcomings.
But price talks, and many commercial customers just want inexpensive, high-speed bandwidth. Business bandwidth providers like Yipes already are delivering these services to paying customers. Extreme, which is in the midst of buying Optranet, a broadband IP services firm, is betting its company that Ethernet really will go everywhere.