How the unlicensed frequency is growing into an industry
By Daniel Sweeney

In a sense it's reminiscent of the earliest days of the narrowband public Internet. A lot of little companies with more courage than capital, and a whole lot of skeptics prepared to dismiss the whole phenomenon out of hand. And the naysayers have undoubtedly had their impact, because so far the big deals and the IPOs have mostly eluded the unlicensed broadband industry.

Today that may be changing, however. At least one communications giant, PSINet has opted to play in the unlicensed arena as ISP, and Time-Warner Telecom admits to conducting experiments, though without making any firm commitment to building infrastructure, at least not yet. In addition, Bell Atlantic has announced that it will use unlicensed wireless bridges for T1 provisioning. Moreover, Adaptive Broadband (formerly California Microwave), long a major provider of wireless infrastructure to major telecom carriers, has recently introduced carrier class equipment for the unlicensed bands. And, perhaps most significantly, the number of ISPs purchasing broadband radios tuned to unlicensed frequencies, according to many industry sources, has grown from a literal handful a year ago to hundreds today.

"Our customer list alone totals more than three hundred companies," claims Tom Walusek, director of service provider accounts for BreezeCom (Carlsbad, CA), a well established manufacturer of wireless LAN equipment which has also developed specialized equipment for the unlicensed carrier and ISP as well as a new division to address the carrier community. "We're not alone in this market," adds Walusek. "The total number of companies involved just in the U.S. may exceed a thousand."

Danny Sipe, director of sales and marketing for C-SPEC's (Dayton, OH) western region, and head of that company's ISP sales program, refuses give a figure for the total number of unlicensed service providers. "They're like ants," he asserts. "They're so many now, and so many new ones coming up that you can't count them. I know of hundreds myself, and I get calls from prospects every day." Incidentally, C-SPEC represents yet another manufacturer which got its start in the wireless LAN business and subsequently developed specialized product for the unlicensed ISP.

Notes Bill Baker of Malibu Networks in Encino, still another equipment manufacturer serving the unlicensed market, "The figure of several hundred sounds right at present. But by the end of the year there may be thousands. You wouldn't believe how many inquiries we receive."

Other industry insiders are more conservative. Don Davis, chief technology officer for NewEraCom, an Ashville, North Carolina based ISP and consultant to ISPs on wireless access, states, "the number of companies really in the business in the U.S. is probably less than a hundred. There may be several hundred companies altogether who've purchased equipment, but most are merely conducting experiments and are not offering regular service."

Davis is quick to add, however, that the growth rate in the unlicensed sector "is phenomenal thus far. Since the first of the year, we've noted a doubling of the number of companies every forty-five days."

Davis's selection of a January '99 date as a convenient reference point for plotting industry growth statistics is significant. As recently as December of last year, Andrew Seybold, the San Jose based publisher of Seybold's Wireless Outlook, and a noted industry consultant in the area of wireless data, indicated in a telephone interview that he believed that carrier activity in the unlicensed band was almost nonexistent. This opinion was seconded by Bill Frezza of Wireless Computing Associates (Yardley, PA), another well known authority in the field of wireless data. The author himself published a survey of the unlicensed industry at the beginning of this year which arrived at similar conclusions.

Since a majority of the dozens of wireless ISPs contacted for this report were already offering service by the beginning of this year, such assessments were clearly wide of the mark. But with no one systematically tracking the industry at the time, and with unlicensed CLECs and ISPs making little attempt, for the most part, to publicize their ventures, it was easy to assume that the industry itself was stillborn.

Even now the unlicensed wireless carrier and Internet access industry still remains largely invisible to the larger telephony and data communications industries, and among the target markets of business and residential users, visibility is even lower. Clearly, unlicensed service providers are going to have to perform a great deal of missionary work before they manage to excite anywhere near the recognition of the rival of CATV data and DSL services. But if the proliferation and growth rates of the pioneering unlicensed firms are any indication, the unlicensed contingent is due to shed its obscurity fairly quickly.

High Growth, High Risk

Of those ISPs offering service today, most claim individual subscriber bases of under 100, notable exceptions being Nobell Communications, Inc. in Austin, Texas, and DirectNet, headquartered in North Lauderdale, Florida, both with a purported three thousand subscribers. Furthermore, almost all current providers are restricted to one or two geographical market areas. PSINet, with its dozen or so operational markets scattered through the South, is presently the only unlicensed ISP of true regional as opposed to local significance.

Neither can the economic impact of unlicensed carriers and ISPs be said to be substantial at present. In every case wireless ISPs strive to price their services well below the cost of a T1 leased line, the local exchange being the key competitor for high speed data services in the secondary and tertiary markets which most of the wireless companies occupy. Cost of service will vary from $50 per month or less up to several hundred dollars of month, and appears more apt to be predicated upon the nature of the local competition than on the quality of service or throughput rate offered by the unlicensed ISP. While return on investment can be fairly rapid, given the fact that subscriber terminals can be had for less than $1000 apiece and access points for under two thousand, the relatively small number of subscribers supporting the services thus far necessarily limits service revenues, and correspondingly the resources these companies can marshal in their competitive struggles.

Simple arithmetic indicates that with a service fee of $200 and a mere hundred persons paying it, the revenue will only amount to $20,000 per month, seemingly only sufficient to meet operating expenses and not nearly enough to fund expansion of the network. And with venture capitalists generally disdaining the unlicensed carriers, the unlicensed would appear to be hard pressed to prevail against rival broadband technologies, which in many cases are represented by companies with relatively vast resources.

Nevertheless, Breezecom's Walusek, for one, is undeterred by such dismal analyses. "It's possible to bootstrap an unlicensed ISP successfully," he insists. "That's what makes this such an exciting business."

And, indeed, a fair number of the carriers are expanding their networks rapidly, including Global Pacific Internet in southern California, the aforementioned PSINet, and Airwire.net and DirectNet in south Florida. Moreover, some, such as WorkNET in St. Louis, Missouri, and Nobell Communications, have announced highly ambitious plans of establishing service in scores of new geographical markets over the course of the next year.

Comments Ken Harrington, president of WorkNET, "everybody's talking about staking a claim in a hundred individual markets. We're talking about it too, and we intend to do it." Nobell's Johansen is similarly ambitious. "We expect to have millions of customers in the not too distant future."

Bill Baker, whose company supplies equipment to many wireless ISPs, sees nothing overreaching about such rapid expansion strategies. "This business is about customer acquisition. The more initial signups, the higher the valuation on the company. In almost every market, the unlicensed carrier will be competing with other access technologies so he has to establish a niche quickly."

Defining the Service Offering

Unlicensed wireless ISPs are well aware that they're not alone in the universe. Actual or potential competitors include the wireline telecom incumbent, wireline ISPs in all their diversity, DSL providers, cable television companies, fiber ring proprietors, public utilities owning dark fiber, satellite carriers, and other wireless services operating in the MMDS, LMDS, cellular and PCS bands.

Thus within the broadband arena, the unlicensed wireless ISP is merely one of many aspirants and scarcely the most prominent in the public eye. And, since, in most cases, these competing broadband services are provided by very large licensees with presences in a multitude of markets and strong revenue streams from established user bases, the situation of the unlicensee--generally a relatively small company--must seem perilous.

So what competitive advantages does the unlicensed ISP bring to the fray to compensate for his fledgling status?

"Lower capital outlay," avers Bill Baker. "By operating over unlicensed frequencies you can become competitive without dishing out significant funds."

Richard Frizalone, a vice president at PSINet, sees other advantages as well. "Time to market is a big plus for wireless in general including unlicensed equipment. But when you look at LMDS and other licensed frequencies, you see a lack of presence on the part of the Winstars and Teligents in secondary markets. Opting to use the unlicensed bands allows you to expand market by market."

Greg Piero, CEO of Wireless Internet Services (Visalia, CA) cites other strengths peculiar to the unlicensed carrier. "You can build and repair infrastructure very quickly, and, provided you choose the right equipment, you can scale the network very cost effectively."

Still, most wireless ISPs appear to be under no illusions regarding the intensity of competition for the broadband customers, and no one is suggesting that wireless, whether unlicensed or licensed, is going to own the broadband marketplace. Most see continuing competition from the LECs offering T1 services, progressively more serious competition from DSL providers, and eventually a formidable presence on the part of MMDS and LMDS licensees and purveyors of fiber.

Most, as a matter of interest, do not voice much concern regarding CATV operators. "Cable companies generally don't pass business districts," Ken Harrington points out, expressing an observation common to the unlicensed. "Small and medium sized businesses are our market," he continues, "while they [CATV] tend to concentrate on residential."

Some, though not all, are similarly dismissive of DSL. "A lot of customers are simply not reachable by DSL," maintains Scott Lingren, vice president of sales for Clearwire in Dallas. "Frequently, the existing copper infrastructure is just too dirty to support the service, and no one's going to lay new copper to provision DSL."

Pete Nicoletti, president and CEO of DirectNet in south Florida, sees DSL providers competing as much with one another as with the wireless contingent. "Northpoint, Covad, and Bell South are all chasing one another's tails around here. DSL is a smorgasbord--ADSL, SDSL, and so on, and you've got the problem of inbundle interference to contend with. Still, you've got Northpoint spending tens of millions of dollars to get into a given market. You can't say they're not serious."

Opinions also differ as to the competitive threat of LMDS and MMDS. Several unlicensed ISPs, including SkyLynx in Fresno, California, and WorkNET, own some MMDS spectrum, and none of our respondents is prepared to discount the potential of MMDS data services. Transmitting near to the 2.4 unlicensed band, MMDS operators enjoy the same wide coverage and extended range as unlicensed operators along with the protection inherent in licensed bandwidth. But LMDS is a different matter.

"LMDS systems are expensive to provision," states Till Fritzsching, president of SkyLynx Communications, expressing a common objection. "The cost issues definitely work against LMDS," agrees Baker. "There's also a problem with scalability, the fact that ranges are so short and so much additional equipment is required as customers are added. But I think the biggest problem facing the LMDS carriers is their basically defensive strategy. Their making a bet that spectrum ownership is a valuable commodity and that the spectrum can be resold profitably. They're just like the first cellular operators in that regard, and that strategy won't build networks."

The Nature of the Networks

In essence much of the activity in the unlicensed spectrum still represents a makeshift, an imaginative redeployment of spectrum and hardware initially intended for very different purposes.

The unlicensed frequencies available for the transmission of data include the original ISM (industrial, scientific, and medical) band at 902-928MHz, the more recent channel allocation at 2.4GHz, and a total of three adjacent bands in the 5GHz region centered at 5.1, 5.2, and 5.8GHz respectively. The 900MHz band, a crowded area of the spectrum rife with interference, is little used in data transmissions today, the principal exception being Metricom, a carrier which is unique in many other respects as well. Most of the current activity occurs instead at 2.4GHz, a band which has been open for several years, and which has seen steadily increasing use in wireless LAN connections for inbuilding systems, campus wide networks, and private wireless bridges--the last consisting of point-to-point connections extending over hundreds of yards or even miles.

The 2.4 band, unlike the 900MHz band, is restricted to data and digital radio telephony and, much to its advantage, is not open to control signals for radio activated devices such as garage door openers. The 5GHz bands are similarly restricted to data, and are even less crowded and thus less subject to interference; however, since they have only recently been allocated by the FCC, equipment is in somewhat short supply, and interoperability standards are not in place.

Further unlicensed spectrum is likely to be made available in the 18GHz region, an area, interestingly enough, where the first commercial wireless LAN system operated, namely, the Motorola Altair. In addition, all transmissions involving free air optical carriers, whether diffuse infrared or coherent lasers, are unlicensed. Since nothing prevents an unlicensed carrier from using more than one unlicensed band, a single system could easily occupy a couple of hundred megahertz of bandwidth, an amount comparable to an MMDS system or to the 24GHz licensed high speed networks currently being established by Teligent.

Unlicensed spectrum is by definition bandwidth that is not for the exclusive use of any one party in a given area. Consequently, a carrier wishing to operate in the unlicensed band need make no applications to the FCC and pay no licensing or usage fees. Unlicensed does not mean unregulated, however. Users are constrained in the amount of radiated power they can transmit in order to limit inband interference, but must tolerate potentially unlimited interference from other users so long as any individual contribution to the total interference does not exceed FCC mandates. The well known phenomenon of heterodyning or beat frequencies also subjects these bands to interference from licensed users in adjacent bands, such extraneous interference being the normal predicament in r.f. environments generally, though in the case of the unlicensed, the out-of-band interference combines with the in-band to create an especially problematic operating environment. For these reasons spread spectrum radio modulation systems with high resistance to interference and fading predominate in the unlicensed space.

Transmissions within the popular 2.4 band are subject to deep fades from multipath interference, that constituting the biggest single environmental constraint in this area of the spectrum. Transmissions in the 5GHz bands are less affected by multipath distortion, but are considerably impeded by physical obstructions such as foliage. Higher frequencies, such as the 18GHz band, become subject to rain fade and other forms of atmospheric interference, and generally require transmission paths which are entirely free from any intervening objects, however small.

In closed, inbuilding and even business campus environments, unlicensed wireless LANs have proven themselves generally reliable, though preliminary r.f. studies are often mandated, and careful installation is a must. Significant interferers are rarely present on the premises, and thus interference when it intrudes, comes from sources outside the area served and is normally considerably attenuated on that account. But a medium area public system serving a metropolitan user population is another matter entirely because interferers are inevitably active within the total footprint of the public network itself. They are in one's midst and impossible to control.

When the first experiments in serving data subscribers over unlicensed frequencies were made in the middle nineties, many questioned that the services could ever be feasible because of the interference issue, but carriers insist that those questions have been largely resolved.

"It's really not an issue anymore," insists NewEraCom's Don Davis. "Our own customers love the service. In three years we've never had anyone drop." Nobell's Johansen makes similar claims. "We have zero churn."

Johansen attributes customer loyalty specifically to network reliability. "We've achieved four nines availability and we're shooting for six. A properly implemented unlicensed system is as reliable as the best licensed network, wireline or wireless." Other CEOs were generally less specific as to availability, though Nicoletti claims three nines availability for his system.

An unlicensed ISP executive who asked not to be identified disputed such optimistic claims. "I think there will always be some problems with wireless reliability, and even moreso when you don't control the spectrum. Obviously, you try to maximize reliability, but you also price aggressively and offer a given rate of throughput at a much lower rate than the wireline incumbent."

Pete Nicoletti of DirectNet has his own spin on that issue. "No one denies that the telephone company offering T1 connectivity is providing a reliable connection. But when you put the local loop charges on top of the Internet access fees, you've doubled the price of service. Take out the loop charges and you've won."

Whatever the problems present in today's fledgling unlicensed networks, except in the case of St. Louis which has two unlicensed ISPs ViperNet and WorkNET (ViperNet executives refused to be interviewed) such networks operate in the absence of direct competition--in other words, no other public carrier is sharing an unlicensed band within a given market. That, however, could change because the absence of protective licenses means that anyone is free to launch a directly competitive service utilizing the same frequencies.

With the explosive growth in service providers, the possibility of eventual clashes is certainly present and acknowledged, albeit reluctantly by some in the first generation. "Sure, you could get two of us transmitting in the same area and interrupting each other's transmissions," says Richard Frizalone of PSINet, "but if someone comes in after to you, you'll interfere with him as much as he interferes with you. What's he gonna gain? He'll just succeed in wrecking both businesses, yours and his."

And Frizalone dismisses the latter as an unlikely eventuality. "I don't think the situation is as bad as some people paint it. Granted you don't have a license to protect you, but if you have any sense you've already occupied the prime locations in the area for your hubs and you've negotiated exclusive agreements so that space cannot be leased to others operating in the same bands. Instead of occupying spectrum, you're occupying the transceiver sites that enable you to use the spectrum. It's a different form of protection."

Beyond basic issues of network availability and reliability, ISPs must contend with the fact that available equipment is IP based (Air Data's (Naples, FL) wireless frame relay equipment and NEC's (Herndon, VA) yet to be released wireless ATM gear are singular exceptions) but will be competing with ATM based LMDS equipment. Questions naturally arise as to how or even if the unlicensed systems will provision the emerging multimedia services such as IP voice, video conferencing, telemedicine, and whiteboarding applications like Microsoft Net Meeting. A handful of equipment manufacturers such as IJNT International are experimenting with reservation protocols that will supposedly enable such applications to run over wireless IP, but, for the most part, the ISPs are content to concentrate on building markets and to worry about specialized service offerings later.

The Comer

Wireless access in the unlicensed bands is a business still seeking to define itself. The sheer number of entrants suggests that the business has enormous potential, but the modest size of most individual metropolitan subscriber bases indicates that that potential has yet to be realized. The fact that so many ISPs double as consultants and attempt to sell their services in the third world is testimony to the highly provisional nature of the business today. Yet if time to market really is the key consideration in dominating broadband access, the unlicensed carriers seem well positioned to succeed. With scant resources and often jury rigged equipment they've already blanketed the United States. What will they do when money and carrier class infrastructure both become readily available?