Special Reports Archive
- Written by Gordon Cook
- Category: Special Reports Archive
- Published: 03 July 2008
- Hits: 29302
A Survey Of The Technologies, The Players And The Prospects - Including An In Depth Look At Metricom
An Introductory OverviewBy next summer wireless as an on-ramp to the Internet is likely to be widely available. This report will examine the major technologies and players. It will then show what some of the services will look like and make some judgments about the impact on the cost of access to the Internet when wireless joins telephone and cable TV as a means of connecting users and by-passing the local loop .
While overall numbers of wireless users are growing, according to some reports, by 40% a year, the numbers are overwhelmingly on the voice and not the data side. Consider the figures published in the July 11, 1994 (p. 16) US News and World Report:
The Minneapolis Star Tribune also emphasized cellular's growth (7/3/94, p. 1D). "As a result of decreased cost and increased convenience, cellular phones are on their way to becoming common accessories of everyday life. No longer a businessman's toy, they are being widely used by consumers, particularly women interested in personal security. The Cellular Telecommunications Industry Association, based in Washington, D.C., said that nearly 6 percent of the U.S. population, or 16 million people, have cellular telephones. The growth rate in subscribers is a mind-boggling 14,000 a day. Jerry Blake, research editor for Radio Communications Report, a Denver-based trade publication, said that in metropolitan areas the usage is higher than the national average, ranging from 9 to 11 percent."
The vast majority of wireless usage is for voice. No one seems to have a good tab on the number of data users but, even if corporate nets like UPS and Fed EX are included, the total number is probably less than 500,000. Nevertheless, many are beginning to suspect that wireless data is where the Internet was three years ago - poised for extremely rapid growth.
Wireless is only now sitting at the end of the runway, poised for takeoff. Until very recently, few people have thought seriously about having wireless replace wireline networks. Why? Because, unlike what is perceived to be the almost limitless carrying capacity of fiber optic networks, wireless technologies use the radio frequency spectrum. Frequency spectrum has been viewed as limited and because of the cost of licensing wireless solutions have been priced higher than wireline alternatives.
Since we are interested in whether wireless technologies can act as tools for further democratizing the Internet by facilitating low cost by-pass of the telephone companies should their prices become too high, we shall focus first on the question of spectrum availability and on tools that allow increased access to spectrum. We shall then survey the basic technologies and players before finally returning to an in depth look at one or two companies which already have the technology and the market strategy that could give the Internet an option in the face any unpleasant surprises from the phone companies. (Disclaimer: writing mostly for a US audience we take a US centric point of view. Opportunities for wireless connectivity should be even greater in the rest of the world than here.)
Does a Spectrum Shortage Inhibit Development of Wireless Data Nets?During the past five years two developments which challenge conventional assumptions about spectrum shortages have occurred. The first is that computers and radios are merging. Civilian applications of spread spectrum radio technology have been developed using techniques that take allow spectrum to be used much more efficiently. The development of digital versions of these same techniques has accelerated this trend. The second is that the FCC, in Part 15.247 of its rule making procedures has allocated portions of the radio spectrum in which these low powered devices can operate without licenses. While some have wondered whether the FCC's doing this signals an intentional admission that spectrum is no longer scarce, most seem to feel that the FCC intended no such conclusions to be drawn. In fact these spectrum sections were never available for conventional communications having been originally set aside for Industrial, Scientific, and Medical applications where radio waves are used to produce heat. The FCC first allowed these uses in May of 1985. Now, almost a decade later, with the development of digital radio technology, applications of a much more serious nature have snuck into these openings in such a way that - driven by lower costs - some think that they may be able to challenge similar applications in licensed parts of the spectrum.
Digital RadiosIn the April 11 1994 issue of Forbes George Gilder writes about Steinbrecher digital radios. These are a special implementation of digital radio technology that could plug some new life into cellular and its related CDPD scheme for moving data by reducing the costs of infrastructure needed to do the job by several orders of magnitude. Gilder points out that the cost performance of analog to digital converters and digital signal processors to which these chips serve as front ends have dropped by as much as four orders of magnitude in the last decade. This means that cheap and distributed computing power can begin to drive digital radios. The implications are enormous.
As Gilder explains it: "In Steinbrecher radios, all of the frequency tuning, all of the modulating and demodulating, all of the channelization, all of the coding and decoding that so embroil the politicians are performed by programmable digital signal processors and can be changed at a base station in real time. Strictly speaking, [most of the] the tuner and modem are not part of the base station radio at all.
The broadband radio in a Steinbrecher base station can send or receive signals to or from any handset or mobile unit operating within its bandwidth (in current cellular systems the full 12.5 megahertz of the band; in PCS, still larger bands of as much as 30 megahertz). All the processing of codes, frequencies, channels and modulations, as well as all special mobile services, can move onto computers attached to the network. Steinbrecher technology thus can open up the spectrum for open and programmable client/server systems like those that now dominate the computer industry."
Bringing computer intelligence into the guts of a digital radio allows it to be incredibly selective in its use of spectrum. For example, "Steinbrecher radio does not begin by tuning to one frequency alone; it gasps every frequency in a particular swath of spectrum." While Steinbrecher radios may be especially good at doing this, other digital radios can achieve similar ends.
Stretching Available SpectrumThe simplest but probably least powerful way of stretching spectrum is technique called Time Division Multiple Access (TDMA). This involves fitting multiple users onto the same frequency of spectrum buy segmenting the times at which they can transmit and weaving these times together to form a continuous transmission stream. It is analogous to multiplexing of data streams over telephone lines. TDMA (as well as FDMA, CDMA- see below) devices are allowed to operate in the unlicensed sections of bandwidth, but at extremely low power in the one mili-watt range. Various spread spectrum devices also take advantage of these three techniques - as a technology it is more powerful, and devices using it are allowed up to one watt of power in the unlicensed spectrum areas. This gives such devices a range of between one kilometer and five miles depending on the kinds of antennas they use -- and some with directional antennas and line of sight to the receiving radio have gone on the order of 50 miles.
Spread spectrum is well described by Ira Brodsky in a July 26 1993 Network World article called "Putting the ether back in Ethernet."
Spread spectrum "is inherently less prone to causing interference" with ordinary radio transmission. The basic idea behind spread spectrum can be traced all the way back to World War II when "Hedy Lamarr and her friend, composer George Antheil, thought up an idea for a jam-proof radio system that would guide torpedoes to their targets. Their idea was fairly simple: Design the transmitter and receiver to hop from frequency to frequency in synchronization. An enemy would have a difficult time jamming a signal that is a moving target. This type of spread spectrum is called Frequency Hopping Spread Spectrum (FH/SS)."
Spread Spectrum and CDMAFrom a more technical point of view we may say that spread spectrum "combines user information with a pseudo random code to create a signal that uses considerably more bandwidth than the user information would require by itself. Spread spectrum "spreads" the signal out over a wide frequency range. In the case of frequency hopping, the spreading is accomplished by serial hopping. The other popular type of spread spectrum is called direct sequence. In Direct Sequence Spread Spectrum (DS/SS), the signal is spread out continuously. In other words, rather than hopping from channel to channel within a band of frequencies, it uses a much greater bandwidth than a frequency hopping system all the time." At first this seems counter intuitive: using more spectrum to have the effect of impacting less spectrum or bandwidth.
However because spread spectrum signals are so spread out, they "look like background, or white, noise to conventional radio signals. Likewise, conventional signals look like white noise to spread spectrum transmission (SST) receivers." As digital signal processors grow in power, they can, in effect, encode meaningful data into what appears only as background noise into transmissions at any given frequency. Brodsky adds "There are other advantages to SST. One that stands out is its ability to overcome perhaps the most pernicious radio impairment --multipath fading. This occurs when a signal takes more than one path to the receiver and arrives out of phase with itself. When the out-of-phase components combine at the receiver, they may cancel each other out. Fading is a frequency- specific effect; at any given time, it occurs only in specific frequencies at given location. Because spread-spectrum signals are spread out over a range of frequencies, they can overcome multipath fading."
But another advance turned out to be possible. Spread spectrum systems could themselves be multiplexed. Given a continuing drop in the prices and increase in the power of digital signal processors, the "coding" behind the generation of the "spreading" in spread spectrum systems could be varied. By using different spreading codes many subtle variants of spread spectrum could be developed and act like "languages." A digital radio broadcasting in one frequency range within a spectrum might speak one language (use one spreading code) and communicate with receivers equipped to understand that language. Other radios coded to a different "language" could broadcast under exactly the same conditions at the same time, on the same swath of frequencies, and in the same place where other spreading codes or "languages:" would be like white noise.
Here is Gilder's description [Forbes, 4/11/94]. In 1991, "Qualcomm unleashed a bombshell. Exploiting the increasing power of DSPs to process digital codes, the company demonstrated a spread spectrum, Code-Division Multiple Access (CDMA) modulation scheme that not only increased capacity some twenty fold over analog but also allowed use of the entire 11.5 megahertz of the cellular bandwidth in every cell. To prevent interference between adjoining cells, analog and TDMA systems could use a frequency in only one cell out of seven."
[An engineer with whom we spoke suggested that the figure of 20 fold was a theoretical maximum which has so far been unachievable in testing. Industry belief is that it is in the 8 to 10 range. If this figure is not significantly exceded some of the advantage of CDMA over TDMA disappears. Note that a CDMA design is more expensive than a TDMA approach to the same design problem.]
Are FFC PCS Auctions White Elephants?In view of these technologies Gilder is a bit apoplectic when it comes to the FCC's intention of auctioning off 2,500 slices of spectrum in the two giga hertz range for 10 year PCS licenses later this year. (For an assessment of PCS - see below.) "So what does this have to do with the impending spectrum auction? Almost everything. Strictly speaking, the FCC is leasing ten year exclusive rights to radiate electromagnetic waves at certain frequencies to deliver PCS. This entire auction concept is tied to thousands of exclusive frequency licenses. It has no place for broadband radios that treat all frequencies alike and offer bandwidth on demand. It has no place for modulation schemes that do not need exclusive spectrum space. Continuing to use interference standards based on analog transmissions that are affected by every passing spray of radiation, FCC rules fail to grasp the far more robust nature of digital on-off codes with error correction. By the time the FCC gets around to selling its 2,500 shards of air, the air will have been radically changed by new technology."
"The FCC is fostering a real estate paradigm for the spectrum. You buy or lease spectrum as you would a spread of land. Once you have your license, you can use it any way you want as long as you don't unduly disturb your neighbors. You rent a stretch of beach and build a wall. The Steinbrecher system, by contrast, suggests a model not of a beach but of an ocean. You can no more lease electromagnetic waves than you can lease ocean waves. Enabled by new technology, this new model is suitable for an information superhighway in the sky. You can use the spectrum as much as you want as long as you don't collide with anyone else or pollute it with high-powered noise or other nuisances. [Forbes 4/11/94]"
The spread spectrum capabilities of the Steinbrecher radios allows them to operate in the three areas of unlicensed usage (900 megahertz, 2.4 and 5.8 gigahertz) approved by the FCC as well as the 2 gigahertz range being proposed for the PCS auctions - not mention in the 225 to 400 megahertz range reserved for government and air force communications. While broadband digital spread spectrum radios can be employed to find unused (in active) space in already licensed spectrum bands, CDMA allows non disruptive entry into bands that are already in active use. [The engineer stated that what Gilder meant by the difference between broadband digital and CDMA was not clear. However if there are narrow band users in the spectrum in which CDMA is functioning, the liklihood off CDMA being overwhelmed is large. Most commercial systems have adopted frequency hoping to avoid this problem.]
Suggesting that it is time for the FCC to call of the PCS auctions, Gilder writes "It is as if Reed Hundt is auctioning off beach front property, with a long list of codicils and regulations and restrictive covenants, while the tide pours in around him and creates new surf everywhere." [Forbes 4/11/94.]
There is a Spectrum Shortage Afterall?In May 1994 Van Nostrand Rheinhold published Using Wireless Communications in Business, a short book by Andrew M. Seybold. This book is a very useful guide to the current state of the wireless art for the person who is otherwise computer and telecom literate. The book practices the text book principal of telling the reader the essential facts, retelling them again and wrapping things up by telling them for a third time. It also has no case study material - no examples of actual uses or users - except at the one and two paragraph level of explaining how a few systems such as the UPS delivery system works. Consequently it is a dreadful bore to read.
Nevertheless, as a basic survey tool, it does an adequate job. And for anyone coming from reading Gilder the contrast in philosophy is stark. On page 2 Seybold writes: "Currently a severe shortage of radio spectrum exists, despite all the advantages in technology. The Federal government is the biggest user of spectrum in the United States and the FCC has no jurisdiction over these frequencies. . . . . Meanwhile the FCC is in the process of reallocating up to 220 MHz of spectrum in the 1.8 to 2.5 GHz range for Emerging Technologies or Personal Communications Services."
Seybold goes on to describe the cumbersome and time consuming rule making procedure for frequency allocation and the complex process for frequency coordination among users of licensed frequencies. He finds that "in reality with the present shortage of frequencies available within each service, the business radio committee will most likely recommend that new users adopt a "shared radio system" of some type." (p.3) He continues: "because the communications industry is so dependent on a very scare resource (radio spectrum), it has put in place a number of different and effective ways of tracking activities within the FCC. . ." (p.4) And finally: "currently not enough bandwidth is available to permit the realization of visions depicted in the videos, nor is enough bandwidth available to enable all to communicate with whom they want when they want."
While Gilder is optimistic about the ability of spread spectrum devices to steer their own courses on the spectrum oceans, Seybold is pessimistic, pointing out the 902-928 MHz band is shared with many different types of users. Some of these users are licensed and therefore protected when it comes to creating or receiving interference from unlicensed users. "In this particular band unlicensed users rank number five in the pecking order - meaning that four groups above them are protected when it comes to interference issues. Also of concern is the emergence of this band as the new portable, in home wireless phone band, (also unlicensed) as well as garage door openers, remote control devices and even low powered walkie-talkies." (p. 109).
Who's Right?We asked on the telecom-reg mail list whether both Gilder and Seybold could possibly be right? Bennett Kobb of New Signals Research (described to us as one of the most astute FCC watchers around) offered an especially perceptive reply:
"Who in the world came up with the idea that spectrum is not scarce? A day does not pass that I'm not confronted with spectrum scarcity. This is a scarce, precious national resource. We should be as jealous of its protection as we are of Yosemite or the Grand Canyon." - FCC Commissioner (and now head of PBS) Ervin S. Duggan
"Let me tell you while there are more services that need spectrum, and parties who want permission to use it, then there are spectrum resources to give to them, the extent of the shortage is to some degree an artifact of which technology is being applied to use the spectrum and the business plans and agendas of suppliers.
As of only a few years ago, Motorola was still publicly advocating that analog FM is the superior technology for mobile communications. FM is not as efficient as other techniques, but Motorola had not developed those techniques and products to the point of entering the marketplace with them. Their business plan and political agenda supported continued use of conventional technology and active opposition to possible alternatives, until Motorola was ready to introduce its own versions. Given their dominance of the market and regulators, these policies translated into shortages of spectrum.
Like many of Earth's resources, we can find better and more conserving ways of using the spectrum resource. A proven way is to lower power for those applications that can get by with lower power. Unlicensed devices use this method. The unlicensed bands should support lots of satisfactory usage, but they are hardly infinite. We have to get smarter in technology (developing systems that are higher in capacity and more resistant to interference), smarter in business (discovering those needs that can be met with low power) and smarter in regulation (allocating more spectrum to unlicensed radio in recognition of its efficiency and value).
The ultimate question for me is whether supply and demand will seek a comfortable level and the majority of users will get the service they request," Kobb concludes.
Finally he reminds us: "Some scholars don't see the end of shortages. Carolyn Marvin, in "Telecommunications Policy and the Pleasure Principle," argues that: "With the emergence of new technologies of communication promising greatly increased channel capacity and interactivity, scarcity of the means for communicating is sometimes dismissed as a thing of the past. So far this promise is largely theoretical. . . . The introduction of new technologies with orders of capacity, flexibility and speed greater than whatever was previously available has never put an end to channel scarcity. The civilizations to which writing and printing first came learned to swallow the increased volume of messages those technologies made possible and to demand more."
Kobb's point is that while spectrum is finite - technology advances can greatly increase our capacity to use more of it more efficiently. However, economic agendas of major technology providers can and will interact to keep spectrum availability tight while new applications keep demand increasing. In short Gilder writes from the technology point of view and oriented towards the future as though the economic motivation and political constraints did not exist. While Seybold writes oriented toward the present and seems determined to give his conservative business oriented readers a somewhat pessimistic point of view focusing on the economic and political dynamics that interact to keep spectrum availability constrained. (Some industry insiders pointed out to us that they consider Seybold to be far more respected than Gilder.)
But How to Cope With the Practical Problems?No less than five entities can use the 902-928 MHz band. In order of priority they are (1); ISM FCC Part 18 -use of radios waves to produce heat rather than for communication-; (2) goverment including naval radar; (3) Automatic vehicle Monitoring (Part 90); (4) Amateur Radio Part 97; and (5) unlicensed spread spectrum. The first four categories of users have priority over spreadspectrum use. Amateur Radio can involve strong transmissions of up to 1500 watts narrowband and 150 watts spread spectrum.
We asked what would happen if an unlicensed spread spectrum user ran into such interference. Joe Chuk, Professor at Kutztown State College replied: If the receiver (of the unlicensed user) is overloaded by a physically proximate licensed user's transmitter (especially if it's using high power -- although amateurs are supposed to use the lowest power necessary for the intended communication), I can see a problem in any other user's attempt to communicate. An overloaded receiver sees signal everywhere in the band. The obvious solution would be to place some physical distance between the offending transmitter (which is assumed to be running within all legal parameters) and the user with the receiving problem.
It would seem that, in the presence of an extremely strong RF field, spread spectrum technology would not be at any particular advantage or disadvantage. My understanding of Frequency Hopping SS systems is that they use a number of frequencies in a predetermined sequence with very short dwell times (< 10 milliseconds) on each frequency. I don't know that the existence of traffic in the part of the spectrum in use has an effect on what particular frequencies are used. An overloaded receiver, however, would see signal from the offending transmitter on all of the frequencies it is trying to use, and communication would be hampered.
One advantage that SS does have over conventional modes is that the "de spreading" process on the receiving end can ignore interfering signals within the portion of the band used for the SS communication. So, as long as the receiver isn't overloaded or desensitized by a strong nearby transmitter, Frequency Hopping SS could well make it through when other modes that are limited to a single channel (at a time) cannot."
To which Dean Heinen of Northwest Technical Services, Yakima Washington added: It is essentially correct that in the presence of an "in band" RF field strong enough to overload or desensitize the front end of a receiver, there is no advantage to spread spectrum technology. This assertion is not based upon which form of spreading is being used, Direct sequence (DS) or frequency hopping (FH). This is based simply on the fact that an overloaded front end places the receiver in a condition that cannot be recovered from, by filtering, or by interactive control of the placement of carriers (in the case of frequency hopping).
Although [a large number] of spread spectrum systems in the 902- 928 MHz. band use a form of direct sequence spreading, there would be a possibility that in a congested band, with many narrow band communications occurring, the frequency hopping method could be superior, if, a form of interactive control were used in the development of the PN code. That is, if an algorithm were placed in the system to detect the carrier frequencies that got "hit" regularly, and make a determination to eliminate those frequencies, and add another in it's place. This would require a "smart PN code modification" on the fly. To my knowledge, this is not a current feature of any of the frequency hopping systems. [We were told by another source that Metricom radios do use this technique.]
And, all of this may be moot, as the percentage of radios operating FH, as opposed to DS are small. . . . . The Cylink, ARS, Cal. Microwave, and others use DS. DS will also be the basis of CDMA for cellular (when deployed). The method of PN code generation used in DS, and the nature of the spreading sequence preclude the frequency hopping type of "interference avoidance". [Other sources with whom we spoke diasagreed with this evaluation, saying that while no one had exact numbers frequency hopping usage is very substantial.]
This places the user in the unfortunate position of attempting to reduce an offensive ( but perhaps legal ) carrier by conventional methods. that is: filters, antenna placement, and directional antennas, capable of rejecting signal in unwanted directions. Interestingly enough, after many installations of spread spectrum equipment, I have not yet encountered this problem. This may be due to the likelihood that amateur transmissions are by nature, concentrated in non-commercial areas (homes), while the SS installations tend to be in commercial areas (businesses, factories, etc.)" [Amateur users with repeaters are like to havethese devices in areas where there would be commercial traffic. However the total numbers appear to be small enough not to cause great problems.]
Heinan concluded: "Currently, the problems I have experienced with SS are out of band RF signals of very high power. Namely, paging transmitters in the 928-944 Mhz area. These are often of very high effective radiated power, often over modulated, and are the systems that most often overload the front end of a SS system. This can usually be corrected by the use of bandpass filters though."
All in all it seems safe to us to conclude that while spread spectrum systems certainly do not offer a risk-free path to any communications utopia, they are viable in the current 902 to 928 MHz bandwidth and can be used effectively in the 2 and 5 GHz ranges where they are also authorized. We shall come back to examine some specific applications of these spread spectrum technologies as a means of telephone company by pass in Part 3. In Parts 1 and 2 we shall present a quick over view of wireless technology. Finally for a summary of the FCC role in the Part 15 .247 rule making see the accompanying side bar. It certainly seems to show that that to the extent that we have a spectrum shortage its is for political and economic reasons. Why? Because although the FCC's final 1985 action on spread spectrum was historic it was but a shadow of what it had sought to do only a year earlier.
How the FCC Moved to Allocate Spread Spectrum to the ISM Bands - Two Points of ViewIn a telephone interview on August 1 Dewayne Hendricks, a life long "ham" and President of Tetherless Access, called our attention to the FCC rule making process that resulted in the Part 15 Regulations covering unlicensed spectrum in 1985. According to Hendricks the proposal that permitted the unlicensed operation of spread spectrum radios grew out a desire of some of the FCC engineering staff to rationalize the licensing process to enable technical innovation to compete with the economic interests that tended to keep available spectrum locked up by allocations to large corporations or government entities. Some staff felt that an FCC prohibition against the use of spread spectrum radio technology for civilian purposes was hindering technology development and transfer with negative implications on American technology competitiveness.
We made some further inquiries and found out that in December 1979 the FCC commissioned the MITRE Corporation to do a report that MITRE completed a year later as Walter C. Scales, "Potential Use of Spread Spectrum Techniques in Non-Government Applications, the MITRE Corporation, PB 81-165284, December 1980. By May of 1981 a Notice of Inquiry (NOI) General Docket Number 81-413 was written. It was adopted on June 30 and released on September 15, 1981.
The NOI (published in Federal Communications Commission Reports, pp. 876-889) stated "This Inquiry is designed to serve two purposes. We hope to gather information to: 1) assist us in identifying specific radio services presently authorized by the Commission, as well as ideas for new services, where the authorization of wide band [spread spectrum] modulation techniques would serve the public interest; and 2) identify the technical parameters which characterize a wide band emission, including procedures used to measure these parameters and identify technical standards necessary to insure operation on a minimum interference basis." (p. 877)
The FCC NOI pointed out that the regulatory approach had been one of "reducing bandwidths to achieve higher spectrum efficiency." It added that "reducing bandwidth is at least one way of increasing spectrum efficiency but it may not always by the best way." It pointed out that although the "military today is developing a number of operational systems, there are few civil applications of this technology." The FCC added that much of the military spread spectrum research had been declassified and that new advances in technology were making spread spectrum techniques much less expensive. "Current FCC rules however implicitly forbid spread spectrum's use in most services. This alone may be inhibiting research and development in civil applications." (p. 878)
"The low power density and interference suppression capabilities of spread spectrum systems suggest a unique application, that of band overlay. It may be possible in some circumstances to overlay spread spectrum systems on spectrum used by conventional services with little or no mutual interference. obviously this would increase the spectrum efficiency of the affected band and could release additional spectrum for allocation to other services." (p. 882)
Another source in the developer community with whom we spoke and who claimed to have done much direct research at the FCC pointed out that a Further Notice of Inquiry and Proposed Rule Making Document on the use of spread spectrum didn't come out until May 24 1985. Finally a year later, in May of 1985, the Part 15.247 rule allowing unlicensed spread spectrum devices powered at one watt or less was issued. This source told us that over the four year period several things had happened to weaken the intent of the 1981 Notice of Inquiry. First that power of up to 75 watts had been considered. Second that big companies including Motorola, RCA, and GE had complained that it might interfere with their licensed operations. Third that the intelligence agencies also sought to restrict it because it would make their job of tracking the communications of the "bad" guys more difficult.
Then on August 2 we interviewed a technically knowledgeable Washington DC source who had a differing view - both of these claims and the idea that Part 15.247 could be seen as an admission by the FCC that was no longer a shortage of spectrum. The source explained that the bands chosen for unlicensed spread spectrum were known as the ISM or Industrial Scientific and Medical bands. These bands date back at least as far as world war two and have been established to dedicate relatively small areas of spectrum where radio waves could be used for heat producing properties (as in microwave ovens and diathermy equipment) rather than for communication.
He observed that the FCC had a dilemma of finding a spectrum allocation to permit civilian development of spread spectrum. Finally the "light" went on as they realized that if they stuck spread spectrum in the ISM band allocations they had two dividends. First a micro wave oven couldn't send a complaint to a commissioner and second the spectrum usage characteristics of spread spectrum would enable such radios to by pass the transmissions of radio devices used to produce heat rather than for communication. Thus it very likely struck the FCC as a win-win situation where spread spectrum devices could be allowed to operate in areas effectively closed off anyway to ordinary radio communications technology. A microwave oven operating in the same area might decrease the range of the spread spectrum radio application but since ones like wireless LANs were under consideration a slight loss of range didn't seem harmful.
There was a view at the FCC that the commissioners ought to be getting out of the way of technology and not just looking at the petitions in their in-boxes and picking the winners and the losers. The FCC thought that by giving spread spectrum a chance before anyone begged for it, its usefulness could most easily be determined in the marketplace. The FCC wanted to avoid the self fulfilling prophecy behind saying openly that spread spectrum was the "right" technology.
The FCC evidently did look at higher power than one watt but concluded that one watt would allow under average conditions communication of a couple of miles and concluded that no one really needed to communicate farther than that without a license. They also pointed out that if they allowed devices of graduated power up to a certain ceiling that the people operating on the higher power would tend to capture the band and you get into an escalation race. One watt seemed a reasonable way to keep the costs down and allow reasonable communication. If the FCC had allowed 70 watts, everyone would have been forced to go to 70 watts and it would have escalated the costs of the equipment. The allowed strength of transmissions was being debated during the full period of rule making from 1981 to 1985.
In retrospect, our source said, given concerns about health effects it seems that the FCC never could have allowed an unlicensed system at that high a power anyway. With an unlicensed device you better be prepared to convince people that it is intrinsically safe. It is rather hard to do this with a 70 watt transmitter. This is as opposed to licensed transmitters where the FCC requires licensees to take upon themselves the legal burdens of safe use.
Intelligence agencies were worried not just about restricting the power of transmission but about the technology altogether. This was a super secret technology not even known outside classified areas until Dixon's book came out in 1976. Legitimate concerns were raised by the agencies between 1981 and 1985. Did they have some degree of impact on what was allowed, we asked? Not much, very little actually was the reply. Both sides were satisfied with the conclusion.
When our developer source saw our August 2 interview quoted above he felt this source was being too politically correct in his answers and sent us the text of the FCC 1984 proposed rule making. [Gen. Docket No. 81-413; FCC 84-169 49 FR 21951 May 24, 1984]
In point 21 of this document the FCC said "we are proposing to allow spread spectrum systems to operate on any range of frequencies above 70 MHz without any restrictions on their occupied bandwidths.
"24. The limits on the effective radiated power from spread spectrum devices, operating on frequencies on or above 70 MHz, are presented in | 15.126(a) (cf. Appendix B). It should be noted that no fixed limits are being placed on the radiated power of spread spectrum devices operating in the 902-928 MHz, 2400-2483.5 MHz and 5725-5875 MHz ISM bands. In these bands, all devices are allowed sufficient power for satisfactory operation, providing they do not cause harmful interference to other users of the bands, or produce unacceptable levels of radiated emissions outside the bands. The proposed rules would authorize spread spectrum systems to share these bands on a secondary non-interference basis with the primary users."
"27. Under the proposed change to Part 90 of the Rules, frequency hopping systems would be allowed to operate on any of the frequencies which are presently available to the Police Radio Service. . . . . However, if the hop rate of these systems is greater than 10 hops per second and more than 10 hopping frequencies are used, then we feel that a maximum output power of 15 watts can be allowed and still not cause objectionable interference to the other users, since the time of continuous occupancy of any single frequency, by the frequency hopping system, will be less than one tenth of a second."
When the final FCC rule making occurred a year later all of these provisions had been removed. Unconstrained by politics or economic considerations the FCC engineers had decided that huge freedom could be given to spread spectrum use. What we were allowed was vastly more limited than what we could have had.
"Mixed" Wireless Technologies - CDPD and PCSThe current wireless market is divided into mobile and stationary segments with voice and or data applications in each. Connecting a user who is traveling in a vehicle into a voice or data network demands a much more complex and expensive technology infrastructure than is needed for a user who is stationary. For this reason most of the wireless market is targeted at the businessman who is traveling, wants to communicate across a wide area network and can afford the expense of doing so.
Voice and Data - CDPDThe cellular phone industry has been plagued by high expense and lack of reasonably priced national connectivity for its users. Prices of about 25 cents a minute go way up when users get hit with "roaming" charges for use of the devices outside of the territory of their "home" provider. Data can be transmitted over cellular using voice channels to have one modem make a direct connect to another for a session. If however such a session is broken before termination - something that is quite likely to happen, the user has to go through an expensive and time consuming process of retransmission. The cellular industry has been looking for a cheaper and more reliable alternative.
In July 1994 several things happened that will change the concerns about fragmented cellular services. First the Justice Department gave final approval to the merger between AT&T and McCaw. Then Air Touch (PAC Bell's California based spin off) and US West's cellular operations merged to form the nations third largest cellular provider. The second largest was created by the July merger of Bell Atlantic and NYNEX's cellular operations. There was much speculation that both of these newly merged RBOC cellular operations would soon become national in scope. The action didn't stop here either. According to a July 24 Bloomberg Business News story: "Nextel Communications Inc. plans to acquire OneComm Corp. for $ 650 million in a bold move to make Specialized Mobile Radio (SMR) companies a viable competitor to cellular."
With Personal Communications Service as a completely digital voice and data system on the horizon, the cellular industry thinks it has found a seamless and more reliable alternative to session oriented transmission in Cellular Digital Packet Data (CDPD). This is a scheme to multiplex data transmission into unused analog channels within cell sites. Consequently it could be seen spectrum-wise at least as a virtually free data overlay of the cellular network.
As Seybold explains it (p. 99) "in any given cell there are multiple radio channels - all of which are not normally in use at the same time. For example if a cell site has a total of 12 channels, it could handle twelve simultan -eous voice calls. The chances of twelve voice calls being in range at any one time are assumed to be slim. If someone within that cell wanted to send data using CDPD, his or her system would request a channel and 'hop' onto it. Since voice has priority, data only occupies a specific channel for small specified periods of time and then 'hops' to another idle channel. The CPDP system can, according to the specification, hop faster than the voice system can put a channel into use."
The CDPD time will be sold more cheaply than voice. Consequently the system is engineered to give voice priority. CDPD is a standards based way of moving data at 19.2 kilobaud over cellular networks. (In a June 8 1992 article in Network World Ira Brodsky pointed out that actual throughput would be considerably less since the protocol overhead was expected to eat up 40 to 50% of the bandwidth.)
The CDPD consortium is composed of McCaw, IBM, GTE, Ameritech, Bell Atlantic, NYNEX, PacBell, Southwestern Bell, and US West. Commercial service started in Las Vegas last November. Earlier this summer McCaw went operational in New York City and Seattle. It has announced intentions of operating in 105 markets by year's end. GTE is beta testing its CDPD service in San Francisco. CDPD however may not be implemented by all cellular carriers and where implemented may not be implemented uniformly. Some observers believe that AT&T's buy out of McCaw means that AT&T is willing to put enough money into CDPD into a national service - if not a standard.
By July of 1994 McCaw was still saying it was "on target" to have CDPD operational in all of its 105 markets by year's end. (Communications Week July 11). McCaw is still almost entirely analog in its cellular systems and as we write is reported to still be 100% committed to TDMA for its digital evolution. Consequently it looks as though McCaw will be a definite source of delivery of CDPD for at least two or three years.
Nevertheless we agree with Seybold's caution on p. 157 of his book: "The reality is that at this time CDPD is still an unproved technology without a single 'owner' that can control its toll out and nation wide implementation. CDPD is being deployed by some cellular carriers, but not all. Furthermore, there does not appear to be a concentrated effort to establish a single user, single bill system to handle differences in specifications and operations from one system to another."
Others also note that CDPD is a technology for carrying digital data over an analog cellular phone system that is expected soon to be challenged by PCS. Moreover the analog cellular phone system will be replaced by digital before too long. When that happens if CDPD is not modified to work on TDMA and CDMA digital spread spectrum cellular systems, it will become an "orphaned" technology. (We have been unable to ascertain the technical feasibility of CDPD modification for use with digital systems, the projected cost, or whether anyone has any serious plans to make the necessary changes. According to what Bill Frezza, an independent wireless consultant, told us: "CDPD only works on analog AMPS systems. It is not designed for use with TDMA or CDMA channels. Note that when an operator upgrades to digital he will always leave some analog channels behind for backward compatibility. CDPD will have to run on these." This is certainly not an optimistic assessment of the longevity of CDPD.)
Digital Cellular on the HorizonIra Brodsky wrote in Network World on September 20 of last year "Here's where things really get interesting. It is a foregone conclusion that the most, if not all, PCS franchises in the U.S-will employ the spread spectrum flavor of digital radio. And to increase their capacity, improve their operating efficiency and offer enhanced services, cellular telephone networks are beginning to migrate to digital, too. But cellular carriers [have been] divided as to which digital radio technology to use.
McCaw has aggressively supported the first digital cellular standard to come along: Time Division Multiple Access (TDMA). The case for a spread-spectrum solution - Code Division Multiple Access (CDMA), developed by San Diego-based QUALCOMM, Inc.- has been so overwhelming that the industry has been forced to adopt CDMA as a second 'standard.' PacTel Cellular, US West New Vector Group, Inc. and Bell Atlantic Mobile Systems, Inc. are among the cellular carriers who have taken the plunge and announced plans to deploy CDMA, and others are clearly leaning that way." However by the end of July 1994 there were 25 TDMA cellular systems deployed in the US and zero CDMA systems.
One of wireless' shortcomings has been its segmentation, if not by geography by technology. Wireless applications are split in one direction by voice or data and in the other by fixed or stationary. No single vendor has dared take on the cost of developing a unified technology that is equally good in all four segments of the market.
The viability of wireless is also being squeezed by economic problems. According to "Wireless Technology" in the May 14 1994 Economist (p. 72) "McCaw's Mr Mechaley argues that the cost of building his firm's CDPD network will be less than 5% of McCaw's investment in its cellphone system, which is America's largest. Better still, he believes, customers will be willing to pay a premium to free their data from the tyranny of the telephone cord. Encouragingly, the revenues of America's cellphone industry over the past decade have amounted to three times its aggregate $14-billion investment."
"Yet, as the cellphone industry well knows, revenues do not equal profits. Thanks to cut-throat competition and recession, the revenue generated by a typical cellular telephone has collapsed from close to $200 a month in 1985 to less than $70 today, leaving the industry desperately unprofitable. Wireless-data firms may be squeezed even harder. To expand their networks nationwide, many will have to apply for 'personal-communications service' licenses in the Federal Communications Commission's forthcoming auctions. That will cost the network operators dear, and rob them of profits for years to come."
A bigger worry is the sheer number of competing networks. Motorola's Mr. Battat, for one, fears that the wireless-data business could swiftly turn into a low-margin commodity business. That is why, he says, Motorola is "constantly reviewing" its investment in Ardis. One of the industry's biggest worries is that the heaviest initial traffic over wireless-data networks will be electronic mail: E- mailing computer nerds are unlikely to be big revenue generators." Such dollar figures juxtaposed with the utilitarian nature of Internet email should give the big hi-tech hi-dollar infrastructure players pause for considerable concern. It may take a very special kind of company and technology to succeed in a commodity business. Let's keep this in mind while turning to PCS as the possible next stage evolution for cellular networks.
Personal Communcations Service (PCS)PCS is intended to be an all digital replacement for cellular voice and data that may replace current cellular technology. According to Seybold (p. 161): "PCS is intended to offer a voice service that will allow a single phone number for a user. When a user is home the system will connect to that user's local PCS unit, and there will be no charges for network services." Unfortunately there are many unknowns within PCS including how it will handle data and most of all the cost of equipment and service.
In the licensing of its spectrum PCS technology is being hit with blatantly political moves that some think could raise its cost enough to sink its marketplace success. The FCC is licensing its spectrum in geographic as well as bandwidth chunks in such a way as to make the emergence on any providers of national scope unlikely. Furthermore, according to Seybold (p.166), two things were unique about the commissions 1993 creation of Personal Communcations Service in the 1.8 to 2.4 GHz band. "The first is that this is the largest amount of spectrum ever released at one time for direct public access. Second, this spectrum is already heavily used by point-to- point microwave systems carrying everything from telemetry to 911 emergency calls.
All these point-to-point systems must be relocated, or the PCS providers must prove they can co-exist on the same band without causing interference. Most of the existing users are willing to move, but the FCC has indicated that the PCS providers must pay the costs associated with such a move. The exact cost of a mass movement of microwave systems to higher frequencies is unknown, but it will be billions of dollars. . ." When we add this to the billions of dollars that the Federal government is hoping to get in auctions for the frequency assignments, we conclude that regulatory moves combined with newer technologies such as Metricom's may render PCS still born.
An unsigned article ("Bell South, GTE Ponder Importance of Data Services on PCS") in the May 23 1994 Mobile Data Report said that neither Bell South nor GTE is exactly sure what users will want from PCS. "The bottom line is that potential PCS licensees - BellSouth and GTE among them - are still wrestling with how to compete against the plethora of existing and future mobile communications networks." GTE has "come to the conclusion that data is very important" for all of the company's mobile services, such as PCS and cellular, says Dr. C.J. Weylan, executive vice president for marketing and business development at GTE Personal Communications Services."
"It's likely that data services over PCS will be developed more quickly than they were for cellular because "that's what the marketplace wants," Weylan says. [But] Why would the market use PCS instead of cellular, Ardis, Ram Mobile Data, specialized mobile radio (SMR) and other voice-and- data and data-only networks available? Weylan says if you assume most people prefer to carry a single device for voice and data, that eliminates the data-only networks (e.g., Ardis, Ram and Metricom)."
"If you further assume a significant percentage of the market wants cellular-quality voice, then you again narrow down the number of competitors, he says. Although Nextel Communications says it can provide high-quality voice, along with other voice and data options through a single phone, "whether they can satisfy a voice user who wants good quality voice is a question," Weylan says. Finally, if you assume the market would want services available with high-speed data rates, then PCS offers attractive possibilities, he says."
If this article presents an accurate sample of market positioning what may be going on with PCS developers is pie-in-the-sky wishful thinking. Completely missing is any sampling of what users would be willing to pay for such wonderful services. The fact that he emphasizes higher speed data rates than cellular when the FCC is only talking 30 (for PCS) as opposed to 25 kilohertz (cellular) channels in major metropolitan areas show him trying to turn a 16 per cent increase in speed into a significant selling point.
After talking about several other proposed applications, it concludes that "there doesn't appear to be much that PCS operators will be able to do that cellular and/or SMR operators will not." Finally it admits that in addition to the heavy costs imposed by the licensing process, the capital costs of PCS networks may be as great as cellular because, although the base station costs will be less, between two and three times more PCS sites will be needed to achieve the same in building penetration.
TeleStrategies Jerry Lucas in his July 1994 Insight newsletter concluded that only AT&T may have deep enough pockets necessary to enter the PCS market. He made it very clear that frequency shortage in cellular bandwidths will not be a driving factor in moving customer into PCS. "The top three cellular markets New York Chicago and Los Angeles are approaching one million subscribers each. 50 MHz per market is adequate for their needs. Market number four, Washington- Baltimore, is approaching 600,000 subscribers with no major problems to date. Market 10, which has about 250,000 subs, obviously has no problem. And all of this is before the introduction of digital which should increase capacity at least by a factor of three using TDMA and likely even higher using CDMA in the future. Year 2000 bottom line: at best, outside the top six markets, the reason a caller is using a two GHz PCS spectrum is because the service is either cheaper, better quality, packaged differently, more private, etc. It is not because the cellular provider turned a customer away due to lack of service capacity in the 800 MHz band."
The Frequency AuctionsFinally while the license process within the FCC seems to have been significantly delayed, no one has any feel how the enormous likely outlays for these licenses will impact the ability of companies to provide service at a cost the market will accept. On May 23 1994, Michael Fitzgerald wrote in Computerworld "Like the scatterbrained white rabbit in Alice in Wonderland, the Federal Communications Commission may be late for a very important date in the future of wireless communications. The FCC was expected to begin issuing licenses for personal communications services (PCS) by May 7, but it appears to have missed its deadline" with FCC officials refusing to comment on whether any pioneer preference licenses had been issued.
Ralph Haller chairman of the FCC's PCS task force, promised that the FCC will hold auctions around mid year for narrow band PCS and at the end of the year for wide band. "Haller noted that commercial interests contributed heavily to the 67 petitions the FCC received for its initial rules on PCS. All the petitions must be considered before the FCC can begin the four-to six-month process of picking an auctioneer for the wideband auctions.
Critics contend that: "the FCC has created a significant number of licensing divisions that will make it difficult to create a nationwide PCS network. It has also required that small businesses and women-and minority-owned firms control a certain amount of licenses. Such conditions do not exist in the cellular realm.
'The FCC has strangled PCS -- this was a great opportunity for the U.S. to really bust open the market for mobile communications,' said Ira Brodsky, president of Datacomm Research Co. in Wilmette, Ill. Brodsky said that 'this thing is so weighed down and divided up' that he would be surprised to see companies make significant investments at the PCS spectrum auctions. FCC officials bridle at these kind of accusations. "We're not talking a couple of years' delay here," Haller said." At the end of July the auctions for ten nationwide services license did finally begin. They have produced larger figures than expected - up from nearly $100 million bid by 27 companies on July 25 to $411.6 million at the end of round 16 on the evening of the 27th, according to a July 28 Associated Press story by Jeanne Aversa.
And then on the evening of July 29 James Sweetman, Evaluator Telecom Issues for the General Accounting Office stated on the Telecom Reg Mail List: "The Nationwide PCS auction is completed. The winners(?) are:
|Market||Bidder (number)||Final Bid|
|N-1 Paging Network of VA||(9065)||$80,000,000|
|N-2 Paging Network of VA||(9605)||$80,000,000|
|N-3 KDM Messaging Co||(5398)||$80,000,000|
|N-4 KDM Messaging Co||(5398)||$80,000,000|
|N-5 Nationwide Wireless Network Corp||(7884)||$80,000,000|
|N-6 Airtouch Paging||(7561)||$47,001,001|
|N-7 BellSouth Wireless||(1006)||$47,505,673|
|N-8 Nationwide Wireless Network Corp||(7884)||$47,500,000|
|N-9 Paging Network of VA||(9065)||$37,000,000|
|N-10Pagemart II, Inc.||(9683)||$38,000,000|
On July 30 Nate Zelnick, Associate Editor, Information and Interactive Services Report added: In case you didn't know, KDM is McCaw Cellular Communications and Nat'l. Wireless Network (now Destineer) is Mtel's former pioneer preference recipient now backed by one Mr. Gates of Redmond. Paging Network of VA is the bidding entity of Paging Network Inc. in Plano Texas. By the way, if 100kHz of spectrum is now worth $80 million, think of how much all currently licensed services cost: RAM is a $300 million company, etc., etc.
Valuable New Services? Or Financial Quagmire?Observers speculated that companies must have services on tap that would go far beyond traditional paging to be willing to offer such exorbitant prices. We found a specific example in the The Dallas Morning News which observed (July 24) that Paging Network (the nation's largest with 3.4 million customers) "has pledged to use the spectrum it might win at the auction to offer a service it calls VoiceNow. The offering will turn a pager into a miniature portable answering machine. Someone who gets beeped will not have to call into a central office to get a message from a receptionist. Instead, the actual spoken message will be played back by the pager itself."
Another winner was Nationwide Wireless - known also as Destineer. According to the same Dallas Morning News article: by 1996, Destineer plans to begin connecting the pager to Apple Newton-style "personal digital assistants' and notebook computers, so that users can send any message they type back by these frequencies. In such cases, the pager will probably look more like a credit card and slip into a slot on the side of the computer or communicator."
"The paging networks then will compete with the cellular telephone networks for the actual transfer of letters, reports, spreadsheets and other files. Destineer's Louise Goodman says its wireless data service will be geared for documents that are the equivalent of five to seven 8 1/2-by-11-inch pages of type. But users can break their larger documents into multiple files, as long as they're willing to pay the freight. Rates have not been established. But basic monthly service fees are likely to fall between basic one-way paging and cellular telephone services, she said."
With roughly 2400 PCS licenses due to be auctioned in the winter of 1994 - 1995 industry observers are now saying that far more than the Clinton Administration's estimates of ten billion dollars will be realized. Observers speculated that bids of up to 80 million dollars for ten years use of some frequency blocks would be combined into larger more powerful blocks in an attempt to offer national cellular phone service.
Are we at the ground floor, poised for the take off of major new industries? Experts point out that the wireless industry is growing at the rate of 40% a year. The smell of a gold rush is heavy in the air. When the press reports that the 10 national paging licenses were expected to generate between 4 and 6 million dollars and they wind up at $37 to 80 million a piece, we are clearly entering unexplored territory.
Jube Shriver writing in the business section of the Los Angeles Times (p. D-3) on July 24 seemed to capture the uncertainties as he wrote: "Despite the promise of the two new technologies, some experts warn of a huge shake out in the wireless industry a few years from now. Prices for existing paging services, they note, have already begun to decline in anticipation of PCS. And following a recent award of wireless services in Britain, one of the victors gave back its license to the government, saying the market was not big enough to support four players, according to Lawrence Hickey, an analyst at First Analysis Corp. in Chicago."
"Once PCS is up and running in the U.S., local markets may be choked with up to 10 wireless competitors apiece: two existing cellular phone providers, seven PCS providers and a new wireless mobile communications service that's being launched by MCI Communications Corp. and Nextel Communications Inc. of Rutherford, N.J."
"When the Lexington, Mass.-based consulting firm Mercer Management asked industry executives this spring how many wireless competitors each market would still have a decade from now, the consensus estimate was three. "A lot of new (wireless) capacity is about to come on line, and there could be some fallout," said R. Preston McAfee, a University of Texas economics professor who is an an expert on auctions. "The key to survival of the industry is going to be finding new uses for" paging services." More than a few ventures are staking their futures on building fields of dreams in the hope that users will arrive.
In the Financial Post (July 23, p. 8) Kelly McParland noted additional skepticism "Herschel Shosteck, a Maryland communications consultant, argues PCS is a figment of marketing imagination, dreamed up by inventors, technicians and financiers 'to rationalize outrageous salaries or their own dreams of wealth. "The industry has come to believe its own propaganda,' says Shosteck. 'I can only explain it in terms of psychological mass hysteria.' "
Wireless Two Way and Email Data Services
ARDIS, Ram Mobile, WYND & RadioMail, Nationwide Wireless and NextelLet's move from the broader context of voice and data services to those specializing in data only - RAM and ARDIS, the new services - WYND and Nationwide Wireless and finally Metricom.
ARDIS is a nationwide data only service first offered by a partnership between IBM and Motorola in the mid 1980s. It was designed to let IBM field service technicians zap data back and forth between customer sites and IBM central offices. Gradually it expanded to serve other field work forces and then anyone wanting general wireless data communications. ARDIS is licensed on a single 900 MHz channel and originally operated via regional 'hubs" in over 300 cities. Roaming was not possible. A user could connect only at the single hub where he was being billed. As ARDIS grew to 40,000 subscribers, it added some additional channels in high density areas and recently made roaming available.
In December 1992 RAM Mobile Data was fired up as a competitor to ARDIS and by mid 1993 was operational in over 100 cities covering 90% of the US population. As Seybold points out in his book (Using Wireless Communications, p.32) "the system is based on the Mobitex network pioneered by Ericsson in cooperation with Sweden's National Communications Authority." Mobiltex protocols are open and public domain. As a result hardware for the systems is available from Motorola, Ericsson and many other vendors. By making use of trunking RAM is able to offer up to 30 channels per service area. Each channel can accommodate between 500 and 1000 users. RAM was designed from the beginning as a nationwide system that offered the capability for seamless roaming.
While the ARDIS system operates in the portion of the spectrum just below that of cellular phones, RAM operates just above the cellular spectrum. Each ARDIS channel has a 25 kilohertz bandwidth while RAM is limited to 12 kilohertz per channel. Since the wider bandwidth the faster one can transmit ARDIS can offer speeds of 19.2 kilobaud where RAM is limited to 9.6. (While ARDIS was limited to 4.8 kilobaud throughput for reasons of equipment, it is now upgrading its network to 19.2 speeds. As Seybold explains it (p. 90) "all the older subscriber units are locked on to the main channel with crystal controlled units" at 4.8 kilobaud. All new ARDIS equipment can switch between the main and newer channels and operate at the higher 19.2 speed. While RAM is ahead of ARDIS in allowing seamless roaming over the entire nation, ARDIS has better coverage in specific areas than RAM.
RadioMailRAM from its inception grasped a key distinction in marketing that ARDIS did not. Another start up company accompanied them at their first press conference in November of 1992 - RadioMail.
Seybold (p.70) describes RadioMail in the following way: "if the wired and wireless communications links are the pipes through which information is sent, RadioMail represents the reservoir and the switching station, or the post office that controls the information put into the pipes. . . . . RadioMail's concept is that it will serve as the one system that will connect to all information suppliers" on the one hand and to wireless data services such as RAM and ARDIS on the other hand. While the wireless services need only to connect to Radio mail, RadioMail in turn connects to MCI mail, ATTmail, Applelink, wire services for news and the Internet. Some users choose RadioMail as a supplement to their normal Internet provider - having it serve as a forwarding system to them when they are on the road. Others use it as their Internet provider. Those in this category have only email connectivity to the Internet and cannot telnet or use any of the gopher or World Wide Web services.
The Marketing ChallengeRAM Mobile Data's original operating philosophy was that it would 'succeed in the wireless messaging market by providing networks optimized for message traffic and by helping developers provide wireless connectivity in their products and services." Seybold finds that RAM has been successful in its approach of giving users access to public as opposed to corporate email systems. "Thus while ARDIS has grown by promoting additional corporate access, RAM has been increasing its market size by enabling individuals to take part in the wireless revolution." (p. 93) "While the population of users is greater than on the ARDIS system, RAM appears to be on the brink of outmarketing ARDIS. ARDIS is beginning to offer RadioMail to its users but RAM has made a deal with Intel that will provide almost instant access to its network for those who want to experience wireless messaging -- without disturbing the corporation's computing environment."
"This is the plan: Intel is marketing and private labeling a RAM compatible wireless modem made by Ericsson. This wireless modem is available in all retail outlets that sell Intel products including its line of networking cards. The potential user has to make only two choices under this purchasing model. The first concerns the mail connection software that will be used. Because of agreements with Lotus, Microsoft, and AT&T, the choices today are cc:Mail (Lotus), Microsoft Mail, or Easy Link (AT&T). The only other choice that needs to be made is whether to run an x.25 line from the [corporate] server to the nearest RAM facility or to use a second wireless modem. The recommended choice is to purchase two wireless modems. Intell sells one that is configured to run from a standard AC power line and has a standard serial port connector for a direct connection to your LAN. This configuration is [so] easy to install [that] the Information Services department should not have to worry about making any changes to its system" (p.95).
Network World recently published comparative pricing information ("Speedy wireless nets" June 20, p.38). RAM Mobile maxes out today at 8K bit/sec, while ARDIS and CDPD run up to 19.2K bit/sec. ARDIS' Personal Messaging service costs $69 a month for 200 messages, while field service personnel average about $100 a month for ARDIS net use, the company said. RAM Mobile messaging costs $135 per user per month for unlimited messaging, though $25 and $75 packages are available that put ceilings on usage. CDPD provider McCaw Cellular Communications, Inc. says users will likely pay an average of $ 35 a month for its service. Metricom, as we shall see, is far cheaper.
Changes in ARDISIn March of this year ARDIS dropped out of the Internet connectivity picture when it dropped RadioMail and initiated "a real-time messaging service, called ARDIS Personal Messaging, based on an application developed by Business Partners Solutions International, Inc. designed for intracompany communications.
Currently, ARDIS relies on business partner RadioMail Corp. to provide users with the messaging application that runs over the ARDIS network and the gateway services to link users to other networks. By offering its own messaging application, ARDIS hopes to lower the cost of wireless messaging by eliminating the third party. However, users will give up connectivity to users not on the ARDIS network. ARDIS officials would not provide pricing details." (Network World, 2/28/94, "Vendors to air products," p.4.
Only a week later Network World (p. 4 ) was reporting on a likely change of ownership at ARDIS. "IBM and Motorola, Inc. stand poised to sign an agreement that will change the ownership structure of the companies' 4-year-old ARDIS Co. joint venture. [The company was formed by Motorola for IBM seven years ago.] Sources close to the deal originally expected IBM to sign a memorandum of understanding last week to buy Motorola's 50% share of the packet radio company for $21 million. On Friday, they said the IBM buy-out is now one of several options on the table, which could include Motorola assuming full ownership, the two companies splitting ARDIS assets or a third-party buyer entering the picture."
"IBM taking control of ARDIS could result in a more cohesive and, ultimately, less expensive service by eliminating a partner that might not agree with its decisions, analysts said. "If I'm a large SNA shop, IBM ownership makes me feel good about the continuing ability of IBM to support my SNA network," said Howard Anderson, president of The Yankee Group, a consultancy in Boston. He explained that since the ARDIS architecture allows it to penetrate deep within buildings, vertical market users, such as repairmen, can use personal communicators to update corporate Systems Network Architecture hosts -- access that is likely to be more of an IBM priority than a Motorola one. The ARDIS network supports 33,000 users today, about 14,000 of which are IBM field service technicians, according to the company."
According to Communications Week the outcome of the negotiations was not announced until the second week in July (p. 5). Motorola would by IBM's share of the company for $35 million leaving IBM dependent on CDPD for wireless data networking. Key to ARDIS' success under Motorola ownership would be the implementation of the Radio Link Data Access Protocol that would enable the network to operate at 19.2 kilobits per second.
New Ventures: NextelWith wireless communications growing at 40% a year three new players have just jumped into the wireless market. "Nextel Communications, (Economist, 5/14/94, p. 72) which at present operates radiophones for fleet vehicles, is planning to transform its existing network into a digital wireless voice-and-data service. Nextel's portfolio of radio licenses offers the potential to provide wireless-data services to two-thirds of America's population. Motorola already owns 20% of the firm; in February MCI, Americas second-largest long-distance telephone company, paid $1.3 billion for a 17% shareholding. So far, however, Nextel's service is available only in Los Angeles, and building a national network will cost it a hefty $2.5 billion.
New Ventures: Nationwide Wireless NetworkAt the end of March, Microsoft announced a $150 million wireless network joint venture with Mobile Telecommunications Technologies Corp. (Mtel), the parent company of SkyTel paging . (Network World, March 28, 1994 p.58) "The Nationwide Wireless Network (NWN) would let users send and receive text messages via portable and personal communicators in the top 300 U.S. markets by the second half of next year. Mtel received a nationwide pioneer's preference spectrum award from the Federal Communications Commission last year, which precludes it from having to gamble for frequency in the agency's upcoming personal communications service spectrum auctions."
"According to David Coursey, editor of the P.C. Letter newsletter in San Mateo, Calif., who has been briefed by Microsoft on NWN plans, an as yet unannounced pager-size communicator with a voice interface will be released in early 1996 from Motorola, Inc. and Wireless Access, Inc. The device would let NWN users issue voice commands that the pager translates into data messages. "This will be a limited voice response capability built into what Microsoft calls a 'pager on steroids,' " Coursey said. The communicator might also sport a touch-sensitive screen, he added."
Perhaps Mtel looked attractive to Microsoft because of its pioneer's preference license? No need to worry. Competitors howled about a tilted playing field and on July 14, 1994 the FCC caved in. According to Jube Shiver writing in the Los AngelesTimes, (July 15, p. D2) the FCC "commissioners, facing yet another court reversal, decided to scrap their controversial plan to give a free operating license to a company that had pioneered a new wireless communications technology called narrow-band personal communications services. Rivals of this company, Jackson, Miss.-based Mtel, are expected to bid millions of dollars for PCS licenses in auctions scheduled to begin in two weeks." Bid they did. Only problem was that with Bill Gates checkbook wide open Nationwide Wireless bid even more -- $127, 500,000 to be precise. Nationwide walked off both with one of the top five licenses (Number 5) at $80 million and with Number 8 for 47.5 million.
New Ventures: WYNDThe June 6, 1994 Mobile Data report announced that Wynd Communications Corp. in San Luis Obispo, Calif., "will likely be the first company to implement Ram's new RAMail software. The DOS- based software offers message delivery to E-mail systems, Group III facsimiles, paging networks and Internet connectivity." These services establish WYND as a competitor to RadioMail.
"Wynd was started by CallAmerica Business Communications, a California long distance reseller that offers services covering the San Jose, Ventura, Fresno and Bakersfield markets. Services also include calling cards, voice mail, paging, call forwarding, electronic mail, Internet access and operator assistance. . . . Wynd Communications (wynd is Old English for "wind," "air," "path" and "gateway") already works with RAM to offer paging services (MDR Nov. 22). Wynd employs WinBeep, a Windows-based pager dispatching program from Fourth Wave, and the Intel Wireless Modem which uses AT commands. The customer uses WinBeep on his computer to enter the paging subscriber's PIN number and to compose and send a message to Wynd's gateway from the Intel radio modem."
Mobile Data Report (6/6/94) concludes that Wynd's "messaging software is similar to RadioMail's and offers such features as "copy," "move," and "print," auto-forwarding, folders for filing messages, replies to messages, multiple address books, options to change the screen colors and a simple text editor with spell-checking. (RadioMail has a bare-bones text editor without spell checking.) Other RAMail capabilities include using software commands to turn off the InfoTAC radio modem, turn the sound on and off, transmit messages to the InfoTAC's LCD, generate an automatic receipt once the InfoTAC receives the message and provide peer- to-peer messaging. Overall, it appears that RAMail is more feature-rich than RadioMail, but neither program offers anywhere near the functionality of landline E-mail software."
Trolls Under the BridgeAny new technology is likely to be accompanied by problems for unwary users. Wireless data transfer is no different. Seybold in talking about RAM, ARDIS, CDPD, and NWN warns that all of these use different parts of the spectrum and therefore demand different radios inside user's modems. He warns (p. 88): "users must first determine which service they are going to use," and adds in boldface all capital letter type: "It will not be possible to move from one to another with the same equipment."
With all the attention being paid to PCS, Jerry Lucas (Insight, July, 1994, p. 2) makes another very vaild point. "there is no lack of PCS standards. There are currently eight under development: IS-54 (TDMA), IS-95 (CDMA), PCS-1900 (GSM), B-CDMA (OKI), PACS (Bellcore), DECT (European), CDMA (Omnipoint) and PHP (Japan). The problem is that each of these is incompatible with the others. The gotcha is that you could easily choose a looser."
Wireless as Local Loop By-pass for Internet Access
Introduction - A New Approach in a Confusing Market?According to the April 4 Computer Reseller News (p.55) "As an increasing number of wireless networks vie for the mobile- computing limelight, the industry is taking a step back to sort out existing and forthcoming technologies." Middlemen are emerging as consultants and systems integrators because no companies offerings seem to fit all the uses a business has and because the choices available are so confusing specialists are needed as guides for the novice.
"All the companies in the wireless market have developed more slowly than they expected, said Richard Shaffer, editor of ComputerLetter, . . . during the Mobile94 trade show last month. It will take at least two years before this tangle starts to unravel, predicted Iain Gillott . . . Link Resources Corp., New York. In the next 18 months, the industry will see more technologies related to cellular digital packet data (CDPD), Specialized Mobile Radio (SMR), satellite and Personal Communications Services and more, Gillott said. Adding to the confusion, vendors need to resolve how to sell the products and approach customers, he added.
Dozens of technologies are currently on the market, in pilot stages or expected this year. . . . . Regardless of recent hype over wireless communications, observers said for wireless networks to succeed, the industry must not view each technology as a winner or loser but, rather, think of what serves as the best practical solution for customers." In other words the pace of technology change is in danger of outrunning the ability people to use the technology effectively.
Metricom - a Fleet Footed Mammal Among Dinosaurs?Into this very fluid mixture is stepping a company with an exciting product that offers a very powerful and cost effective approach to internet connectivity -- one that by-passes the telephone company at the user end. The company is Metricom, a Silicon Valley firm that has been in business since 1985, has a mesh network of spread spectrum radios (902 to 928 MHz) running on unlicensed frequency, and avoids all centralized and expensive infrastructure to give flat rate very high speed data network connectivity to customers within its coverage area.
Metricom uses $700, 13 ounce "toaster-size" radio base stations that can be mounted on telephone poles and stuck to the sides of buildings. It developed its wireless mesh network to do utility distribution management on behalf of California Edison in the mid 1980s. Metricom explains its inexpensive infrastructure by pointing out that data over spread spectrum is easier and less expensive to send than voice because it is not nearly as affected by cross talk.
Metricom's President Bob Dilworth, who put Zenith Data Systems on the map by selling PCs to college students, is known for effective marketing of solid but powerful low cost technology. He explained Metricom's network architecture in the following way in a Newsbytes interview published on June 16 1994.
"About that mesh network. 'There are three kinds of networks you can build. One is a bus network, like Ethernet -- everyone is on a shared line. Another is a star network, like wireless telephones. In a star you have a master - slave where everyone goes through the master." The mesh is something new. "All our intelligence is equally distributed -- every node is equal. It's a giant flat network, and we've developed automatic routing systems.'"
"He continued: 'We hang a radio every square mile. When we hang it we give each radio a geographic address, longitude and latitude. When that radio turns on it looks for neighbors, and exchanges geographic data. Now it knows where all nearby radios are. When you're sending a message across the city, since we can only use one watt of power' on those unlicensed frequencies, 'that has to hop several times to get where it wants to go. The first guy who gets it sees it's for a distant address, and sends it in that direction. It moves across the city, hopping radios -- we have a patent on that concept.'" (Metricom's concept reminds the COOK Report of the way packets are routed across the connectionless Internet.)
Newsbytes continues: "Dilworth says an average city could have as many as 30-40 access points where the wireless radio network hands-off calls to wired networks. "We'd expect you to go on a wireless network and hop 3 - 4 times, then drop to a wired access point. We'd interface those wired access points at the frame relay level," using digital lines that start at 56,000 bits-per-second (bps) and run as fast as 1.544 million bps."
Metrocom calls its technology a Micro-Cellular Digital Network, noting that its mesh architecture can also route packets past busy or non functioning radios. As Business Wire pointed out on June 13, 1994 "Metricom's toaster-size packet radios can be installed on buildings or atop telephone poles, eliminating the costly "tower"- based wireless infrastructures other vendors have built, as well as simplifying maintenance. And unlike other wireless networks, Ricochet [the name given by Metricom to its network] can be easily reconfigured or expanded simply by adding or moving these radio building blocks."
Ricochet was commercially launched on June 28 following "successful beta testing at sites that included Airsoft, Apple Computer, CityNet (Cupertino), Compaq Computer, Hewlett-Packard, Microsoft, Lotus Development, Stanford and Visa." The launch opened a network of nearly 500 base stations "with a coverage area that includes the cities of Cupertino, Santa Clara, Los Altos, Sunnyvale, Los Gatos, Mountain View, Campbell, Saratoga, parts of San Jose and the Stanford University campus."
Ricochet will be in commercial use in at least three additional metropolitan areas, including Seattle, Houston and Boston, by the end of 1994. Metricom expects to have Ricochet networks operating as subscriber services in some 30 major cities, effectively achieving nationwide coverage, by late 1996," according to Business Wire (6/13/94).
The purpose of the Ricochet Micro Cellular Data Network is to allow desktop and notebook computers and Personal Digital Assistants (PDAs) to transmit data locally in a campus or metropolitan area network environment or "across thousands of miles using a radio modem the size of a TV remote control unit. When fully deployed, Ricochet will consist of a web of regional data networks that transmit data wirelessly within a business campus or metropolitan area and, for long-distance communication, can tap into virtually any public or private wired network, including corporate LANs and electronic-mail systems, on-line services and the global Internet."
Previous Economic Limitations of WirelessMetricom points out that acceptance of wireless data networking has faced two obstacles - high cost and relatively low speed compared to wireline alternatives. As a result most wireless applications have been limited to short text based files. While vendors have claimed that the convenience of wireless justifies the premium price, Metricom is suggesting that because it takes advantage of unlicensed spectrum, and use an architecture that enables an extremely low cost capital investment, it can deliver a premium high speed service at a bargain price. (Indeed their communications speed at 77 kilobaud between base stations and modems, and about two thirds of this into the user's CPU is between three and four times that delivered by their competitors. This kind of speed will allow just about all of the corporate desk top applications to run on the Metricom network.)
Some critics have complained about Ricochet's inability to handle data transmission from a rapidly moving vehicle. "Ricochet builds routing addresses on the fly based on latitude and longitude of the user. When a user is handed off to another cell, the routing table must be re-built, which can result in dropped packets, according to Brian Salisbury, president of Metricom's wireless service division. This issue could be addressed by using Global Positioning Receivers installed in the modems." (p. 44 Infoworld June 13, 1994) An unanswered question from the Infoworld article is whether Metricom would ever need to go to this trouble. We think not.
Metricom doesn't see technologically more complex and therefore expensive service to users while they are in rapid transit between two places as part of their market niche. Metricom's President, Bob Dilworth says Metricom is taking a campus like approach for users who are "locally mobile." Why? Because according to BIS CAP International, a Norwood, Mass.-based research firm, only 13% of the 27 million U.S. mobile workers today leave their home base metropolitan area. (See Network World June 20, 1994, "Speedy Wireless net. . ." p. 38) Metricom users can communicate mobily if they are on foot using PDAs because their movement is slow enough for the network to track their changing location. Two, Metricom users can also communicate with each other without even entering the mesh network if they are in direct line of sight with each other. Metricom is making what seems to be a reasonable assumption that if a user works within its service area it has so many other advantages of price and speed that the user will be quite willing to do data transfers from somewhere else than a moving car.
Financial StrategyAccording to Steve Hamm in PC Week (April 4, 1994, p..A7) "Metricom plans to limit its outlays by blanketing high-traffic areas first, then building out as demand grows. 'We watched RAM [Mobile Data] spend $ 400 million to build a national network to serve only a handful of customers, and we learned a lesson,' says Brian Salisbury, who became head of the Ricochet network in January after serving as president of Bell Ardis, a wireless network in Canada. He and Dilworth hope they can finish the gradual build-out for less than $100 million."
"This go-slow expansion plan dovetails nicely with Dilworth's campus marketing concept. They're targeting not just college campuses, but corporations, government agencies, and other facilities that have a campus like layout. Salisbury [mentions again] a study of work habits by BIS Strategic Decisions showing 85 percent of work travel takes place within a few miles of people's offices. Those close-to- home travelers will be Metricom's core customers." "But what of the road warrior who wakes up in Cleveland, lunches in Miami, and works late in L.A.? Metricom expects to complete a nationwide network in a few years to serve them. And, in the meantime, it may arrange for another network to handle nationwide messaging for its subscribers. Dilworth wouldn't comment further, but Ardis VP Rob Euler says he's discussed that possibility with Metricom." Hamm notes that Metricom is a nine year old company and that it so far has never been profitable. However, he adds that with 76 million raised from a secondary stock offering at the end of March 94 Metricom has over $100 million in the bank to finance its build out of its Ricochet network technology.
"Metricom grew out of the electric utility business, providing remote metering devices and services for Southern California Edison. Its Utilinet division, which has branched out into industries including waste-water management and natural gas drilling, brings in revenue of about $20 million annually and is still Metricom's main source of income." (Information Week, 5/30/94, p. 93).
Pricing & Service OptionsOn June 28 1994 Business Wire published the following information (which we have updated) about Ricochet: Service, it said, is available in four performance and price levels. Service fees are flat rate, per month charges regardless of network usage and the amount of data transmitted.
The service levels and pricing are:
|Service||Network Access Speed||Service Access Fee|
|Economy||2.4 Kbps||$2.95 per month|
Subscribers to the Standard, Executive and Premier plans may rent a Ricochet portable wireless modem for $20 per month. Alternatively, subscribers to any service level may purchase modems for $495. There is also a one-time activation fee of $95.
Additional services available for Standard, Executive and Premier customers include access to the public switched telephone network for a flat fee of $5 per month and access to the Internet for a flat fee of $15 per month. [This is really just deliver to an internet service provider log in prompt. The user must must set up and pay an internet service provider separately for his account - see below.]
As an article in the June 26 (C-5) San Francisco Examiner noted, "at $30, even Metricom's highest flat monthly fee is less expensive than sending a dozen or so messages a day on Bell Atlantic Mobile's 2 month-old wireless network for $ 49 a month."
Business Wire (6/28/94) also pointed out that the Ricochet network permits customers who use a home or portable computer to achieve wireless access to their desktop computer. With two Ricochet portable wireless modems and appropriate application software, subscribers have wireless access to all the data on their office computer. Customers who wish to take advantage of this virtual office solution may purchase or rent the modems at standard rates and are offered network access for the second modem at 50 percent off the applicable single subscriber rate.
Internet AccessMetricom is offering wireless Internet access through the Ricochet network on a trial basis to a limited number of subscribers. When it communicates through a Ricochet portable wireless modem, a subscriber's computer will act as a PPP node on the Internet - something that neither RAM Mobile Data nor Ardis has achieved with Radio Mail.
Moreover given prices of $135 a month for unlimited email messaging from Radio Mail, Metricom's apparent $50 a month for the equivalent of unlimited v-fast dial up modem access to the entire Internet including the capability to use Mosaic as a browser for such information resources as Gopher and the World Wide Web is an achievement unmatched, even by most wireline internet service providers. Unfortunately these conclusions fell into the too good to be true category as we continued our reporting of this issue. As we point out below, the $50 a month fee does not include the all important Internet ID and mail box. That must be obtained by the metricom user from the service provider.
Metricom seems somehow Internet savvy in a way that the rest of the wireless industry is not. It has gone to great lengths to obtain connectivity to everything and everyone. Some of the early wireless providers assumed that they needed to allow their customers to talk only with each other. Metricom is ready to launch itself from the current mindset that says strength is obtained from tearing down every possible barrier to connectivity.
The May 23 1994 Mobile Data Report emphasized this aspect of the Metricom story. "Drawing on its own experience and the experiences of Ram and Ardis, Metricom is making a special effort to ensure its network is compatible with AT software. [In other words, to make certain that its users can use the standard ATDT protocols to connect with PSTN attached wireline modems as well as with the wireless network.] Existing AT-based programs that Metricom has tested, such as Lotus cc:Mail, Newton Mail and ProComm, should work about as efficiently over Metricom's network as over phone lines, says Brian Salisbury. 'The key is trying to make your wireless modem in all respects work like a wired modem," he says.
"Subscribers to Metricom who want to use AT- based software will have to either connect a radio modem to their host computer or connect through a gateway radio modem which dials a landline phone number. Because Metricom's network is "peer-to-peer mesh as opposed to a star (configuration) with a centralized message switch, the idea of interfacing to the telephone network has been a feature from day one," says Salisbury. Metricom is installing wireline modems (or "bridges") connecting its base station radio modems to wireline modems at selected sites. To access these base station modems for connecting to an on-line service, a Metricom subscriber would have to know the area code where he's located and the local access number of the service."
"For example, to dial America Online or CompuServe from Los Gatos, Calif., (Metricom's headquarters), a subscriber would enter the area code and asterisk - 408* - and the on-line service's access phone number. Although travelers might not know the area code at their location, it's certainly easy for them to find out, says Salisbury. Eventually, through agreements with on-line services, Metricom hopes to eliminate the need for subscribers to know area codes by connecting base station modems directly to on-line services. Obviously, this is dependent upon customer demand," Mobile Date Report concludes.
Metricom Beta Test Applications at City Net and AppleOn June 20 Mobile Data Report continued its heavy coverage of Metricom with an article on beta test applications. "When Metricom officially launches its Ricochet "Micro Cellular Data Network" on June 28, residents in Cupertino, Calif., will be able to use the network to access CityNet, an on-line service that offers information from the city government, schools, stores, non-profit organizations and churches. Even more important to CityNet would be the ability for students and teachers to access school software via Metricom, but that's not yet possible, says Wally Dean, chief executive officer of CityNet and mayor pro tem of Cupertino."
"CityNet is a non-profit organization and has offered landline access to its databases for 18 months, says Dean. The organization has beta tested Metricom's spread spectrum system for about four months. CityNet's offerings include electronic mail, information about the city's services and meetings, applications for city permits, church schedules, job banks, school announcements and retail store promotions. 'If you want to take out a dog license,' Dean says, you can fill out a form electronically or print it out. Residents can speak out on the open forum dealing with Cooperation government actions."
"Apple Computer's Advanced Technology Group, for example, has been talking to Metricom for more than a year and playing around with the technology, says Susan Schuman, the Newton product line manager for communications products." Apple employees have successfully used the line of sight communications capabilities to exchange hand written messages using Newtons on the Apple Cupertino campus.
Visa Signs Metricom DealOn May 9, 1994 Computerworld (p.60) announced that Visa International, Inc. said it "will work with Metricom, Inc., a relatively unknown wireless vendor, to develop targeted wireless networks. . . . Visa plans to use Metricom's campus-type micro cellular networks to cover areas with little or poor telephone service, such as Latin America, and feed into VisaNet 2000, Visa's worldwide network. 'Wireless offers an opportunity for Visa to improve electronic transactions at the point of sale [and] make it faster and more accurate,' said Dennis Moser, senior vice president of acquirer systems at Visa in Redwood City, Calif.
According to Moser "wireless networks such as Metricom's should allow Visa to provide better customer service in foreign countries with poorly developed telephone systems. For example, in some countries a ceiling is placed on credit card purchases because telephone authorization is next to impossible." Visa plans to add Metricom to its network in 1995. "To do this, it needs to have Metricom-capable point-of-sale terminals in place. One such company, VeriFone, Inc. in Redwood City, Calif., has demonstrated a Metricom-capable network and could deploy its terminals whenever Visa wants, said Joseph Savage, vice president of the network systems business at VeriFone."
It seems to us that Metricom may also have an excellent opportunity in the Russian city of Iaroslavl, 180 miles northeast of Moscow. Only six weeks ago, George Soros's International Science Foundation has committed itself to build a "civic" computer network that is connected to the worldwide Internet. The Foundation intends to connect the local university, three institutes of the Russian Academy of Sciences, the radio station, the TV station, a museum, a mosque, four monasteries, and 100 K-12 schools. It hopes to complete the work between October of this year and May of next year. When we informed a Russian friend of their plans, he snorted somewhat derisively and said first they will have to rebuild the public switched telephone network of Iaroslavl -- that will take them a lot more than 6 months! The answer seems obvious. A Metricom mesh backbone network connecting each institution to every other insititution. Wireline dial up terminal servers at the institutions, and a satellite trunk to Moscow. The Foundation has $5 million set aside for doing this in Iaroslavl and four other cities as quickly as possible. If it is successful in these test beds, it is talking about replicating such networks in Russia's 30 largest cities!
As the Computerworld article concluded: "Metricom President Robert Dilworth said the company is negotiating in the UK, Hong Kong, Germany and South America and is looking to sign up local partners. Dilworth said that Metricom's networks can potentially be less expensive than dedicated telephone lines."
Metricom Outsources Billing to Perot Data SystemsOn June 6, Mobile Data Report announced that Metricom, was "outsourcing its 'back office' systems to Perot Systems. Perot also is exploring with Metricom the possibilities of constructing a spread spectrum-based network in Great Britain. Customer support, billing and telemarketing systems are Metricom's first priority."
"Perot, has evaluated Metricom's technology and strategies for several months (MDR Feb. 14)." Metricom, accepted an offer from Perot to could provide the needed services in time for the company's commercial launch. Metricom also found the Perot arrangement advantageous, because, a a startup network, it isn't sure how much capability is needed for the back office services. Because other mobile data networks, such as Ardis and Ram Mobile Data, started several years ago, it's difficult for Metricom to base its requirements on their startup experiences Perot can scale up or down to meet Metricom's and its customers' requirements."
"At the same time as Perot is configuring billing and customer support services to quickly deploy its first regional mobile data system, the Dallas -based systems integrator also is helping Metricom with long-term plans. [While] Metricom already has in-house billing and customer support for its private mobile data networks, primarily for utility companies. Metricom and Perot are determining whether Metricom's existing computer system should handle those services for both the private and new public networks, whether there should be separate systems or whether Perot should continue to provide some or all of these services."
"Metricom would like to have customers for mobile data networks. Some of Perot's existing customers in the United Kingdom are 'strongly interested' in Metricom's services, says Bill Shaw, who is exploring alliances with mobile data players for Perot. However, the 902-928 megahertz frequencies, which don't require licensing in the United States, aren't available in the United Kingdom, says [Metricom's Brian] Salisbury. Perot, which is "quite active" in Great Britain, is investigating which frequencies are available. Frequencies above the one gigahertz range could be available, but these higher frequencies don't travel as far so Metricom would have to install more base station radios."
August 3, 1994 Interview with Metricom President Bob DilworthCOOK Report: Is there anything in general you'd like to say about Metricom and where you are going having read our draft description of what you are doing?
Dilworth: It's real simple. Wireless communication has cost more and delivered less performance than wireline. We are giving performance that is similar to wired and trying to price our services at about what you would have to pay for wired. We argue that once this happens the market for wireless will be much bigger.
COOK Report: Who do you think doesn't want you to succeed and what if anything are you worried about?
Dilworth: I think the industry is new enough so that the legitimate players in the industry are all looking for successes. I really don't see us getting stomped on by any competitors.
COOK Report: What do you mean by legitimate players? Dilworth: We have seen some small operations that have had some people out without product -- in effect selling vapor. And some get out there and start to talk about products that violate the laws of physics. As I look at real live competitors, I think RAM is anxious for us to succeed. I think ARDIS would like to see us succeed because when we succeed the entire market grows significantly and everyone has more opportunity.
COOK Report: We have seen recent press statements that the wireless market is growing at the rate of 40% per year. Do you agree with that?
Dilworth: I don't know. I have a hard time believing that it is that big.
COOK Report: At one point we believe that Seybold in his book says that at some point a catch 22 for Metricom might be is that folk who are threatened by you put pressure on the FCC to change the rules under which you operate. Do you see any dangers there? It takes the FCC a long time to change these rules does it not?
Dilworth: Yes it would take any such attack quite a bit of time. Certainly with any administration similar to what we have in power today, it would difficult to derail someone who is trying to bring low cost communcations to the masses. When we have meetings at the FCC I tend to wrap myself in the flag and talk about how I am trying to provide a three dollar a month data phone line, and about how we can help save millions of dollars a year in less fuel burned commuting. So I guess I haven't had any real concern there.
[One of those who reviewed this paper commented: "Believe you me, it is taking a lot less time for the FCC to do things than before. I notice that Dilworth didn't mention the pending AVM/LMS matter which can kill the L band for all of us unlicensed SS companies. That could come down before the end of this year. Ask Bob why he is spending so much money with Dick Wiley's communications law firm if he has no real concerns about the FCC.] COOK Report: We understand why you start in campus or urban or suburban environments. But how feasible is Metricom service in rural areas?
Dilworth: Without a cluster of customers Metricom doesn't make sense. But the cluster can be very small. It can be eight or ten users. You are trying to pay for the infrastructure. We are a last mile means of communication and need a mesh made up of a few radios to support it. Therefore we are not the way to hit an individual site in the middle of the desert. I think the guy in the desert and on a mountain top will be LEO's market (low earth orbiting satellites).
Lets say you are trying to process credit cards at a ski resort and you have 15 or 20 terminals there. You could connect those 20 terminals to a vsat or a leased line via metricom network. You could also put a dozen metricom radios into a pueblo and again use the vsat back to the big city.
COOK Report: So the fifteen or 20 different merchants might transmit their data to a mesh of 5 or 6 radios where one radio would in turn do the hand off to the vasat? What about also daisy chaining the radios every couple of miles along the highway back into town? Could you do this?
Dilworth: Not advisable. Our radios don't work very well in a daisy chain.
COOK Report: Can we turn to the question of wireless access to the internet? How do you get the connectivity? How do we get plugged in?
Dilworth: You can connect to an internet bridge our wireless network and the internet bridge would then connect directly to BARRnet. You would be a BARRnet customer. And you'd pay us $15 in addition to your regular $30 a month bill for us to deliver you to their door at 38.4 kbs. Right now you can dial up to a modem pool at 14.4 or for a lot more money get a 56kbs leased line.
COOK Report: Please describe the internet bridge in more detail. Dillworth: OK There are a number of our radios that are hooked into an ethernet. On this ethernet is a Sun workstation that is hardwired to the Internet via a leased line to BARRnet.
COOK Report: If I am a customer at Apple and I use you to access both the Apple Corporate network and the Internet, what happens?
Dilworth: We expect you to use your own software and you would dial in a string of numbers that would take you to the Metricom Internet gateway. We are not ourselves full blown internet providers. In fact Bay Area Internet providers will add our services. You can connect to the Metricom network by dialing one three digit number to connect to the radio acting as your Metricom corporate gateway and a different 3 digit number to connect to the Metricom internet gateway where in turn you connect to BARRnet or Little Garden or whomever you are buying your Internet service from.
COOK Report: Do you have any commercial customers you can talk about?
Dilworth: I'll just give you some hints. Lots of interest from commercial real estate brokers and fire departments. We are seeing heavy sales of our "twofer" deal -- that is people who buy a connection between their home computer and their desk top machine at their office -- two radios, one on their laptop and one on their desk top.
COOK Report: How much traffic stays within your wireless network in the Bay Area and how much goes outside into wireline nets and the PSTN?
Dilworth: We are just getting data on that, I can't give you a good answer yet.
COOK Report: You are supposedly going to go all out to open up new networks in Seattle, Houston and Boston by the end of the year, right?
Dilworth: Well that's not clear. We are going to go where the customers are.
COOK Report: But press reports say you have plans for those three cities where you have beta programs with Microsoft, Compaq Computer and Lotus.
Dilworth: Yeah, we have some customers in those cities but we have got some other places that it looks like we are going to early on. Detroit for example.
COOK Report: Do you have any time line at this point for opening commercial services at places other than where you are in the Bay Area now?
Dilworth: What we are saying is if someone calls up and wants a clustered network anywhere in the country and if they have anywhere close to enough customers to support the network, we'll do the install.
COOK Report: In other words you must have a model and a spread sheet into which you could plug figures and tell someone what the economics would be?
Dilworth: That's right.
COOK Report: Have you seen the latest controversy about the digital telephony legislation that would require networks to be made wiretap ready?
Dilworth: No. We are a packet switched network. It is diffciult to tap into our data because everything is packetized and sent in different routes over the network. But if you are close to the sender or the receiver, you could possible suck up a large number of the packets.
COOK Report: We saw a press coment that you are considering bidding on some licensed frequencies in the FCC auctions. Is this correct?
Dilworth: Yes. We consider ourselves expert at using shared frequencies. So we would be either a potential partner with someone else who was planning on using the spectrum. In such a case our partner could be doing voice while we moved data.
COOK Report: Do you have any anticipation on when a relationship with Microsoft (which of course is a significant investor in you) that would have access to your network built into some of their software might be clarified?
Dilworth: It's not appropriate for me to answer that.
BARRnet and MetricomA west coast user told us: BARRnet's "rack rate" for 14.4 dial-up SLIP is $200.00 per month. This is for unlimited usage; the price includes a modem. (BARRnet does not offer a metered usage plan for dial- up.) It gives you a single-user POP mail account and primary DNS setup. One-time set-up charges are $1,300. [We tried to verify these figures with BARRnet via phone but on our initial calls were unable to get anyone there who could tell us what BARRnet charged. Vince Fuller told us he was technical and hadn't a clue about the marketing side.]
Suppose that BARRnet waived the set-up fee for Metricom users. Suppose further that they instituted a metered price for Metricom users that was closer to prevailing market rates -- say, $30/month with 30 hours of usage, and $2/hour after the first 30 hours. (I can't imagine them going lower than that, or lower than $150/month for unlimited usage.)
Then what advantage would the user have using Metricom rather than a standard dial-up connection?
1) Faster speeds than 14.4 dial-up, but not [significantly] faster than a V.32 28.8 Kbps modem. (28.8 modems cost around $500-600, but so does the Metricom subscriber unit. And I'll forecast that the 28.8 modems will be down to $200 within 18-24 months.
2) Untethered access (within a limited geographical area), but so what? Will I be willing to pay $45/month for the privilege of doing my FTPs from the city park, rather than my office?
The Metricom Modems: Some Technical DetailsFrom one of our sources we received the following:
"Was at the experimenters' meeting with "John" X. They are renting/selling the modems. I talked to "John" there about tech details. Part 15 radios - 902-928. They use the whole band by FH over 162 channels. Radio is the size of a slim HT with 3.2" 1/4 wave antenna attached. 1/3 of radio is NiMH battery.
Uses RS-232 serial connector, appears as Hayes compatible modem to computer. DTE rate is autobaud up to 38.4K, manually settable up to 115K. I requested that in the next ROM release they do up to 115K autobaud. They are in the process of getting a new ROM done which will expand addressing space and provide access controls not now in place. Also, they sell different grades of service, from 2400 up to the max. bandwidth. "John" said that even the 2400 user can get high bandwidth but he has lower priority than "premium" service user when passing data over the network.
On-air data rate is 77 Kbps. "John" said it was a compromise between distance and speed. Higher speed = shorter distance. The modems are full of goodies. I have been hacking on one tonight. Following my note is output from a hack session. Firmware goodies. The modems I have (pal rented them) are at MCDN (MicroCellular Data Network) addresses 01-09XXX (mine) and other modem is 01-09XXX (in Cupertino). I am on the fringe coverage area, just north of Cupertino. I still can acquire a carrier from one poletop in Cupertino. The poletop units look like an upside-down box celular antenna (5/8 wave?)
Latency is pretty high, from 100ms up to seconds. Seems to vary a lot. Half-duplex of course. All things considered the modems work well. They automatically know whether to talk to another address through the network or direct, depending on whether they can hear the other modem.
If I want to connect to the Internet with PPP I type "atd intcup1**ppp" which means connect me to the Cupertino 1 Internet port with PPP. To access the PSTN (dialout from radio to data modems on PSTN) I use another form "atd 408354*7730768", i.e. to dial 408-773-0768. They do not allow LD calls to be made from the PSTN dialouts.
Radio seems to be solidly made, small (!) and charges battery while it operates. The battery is rated 6v 1100 mAH."
ConclusionFor those of us who worry about the impact on the Internet of 'big" government and the "big" telephone companies Metricom is certainly the most hopeful entity on the horizon. Two of the most important things going for it are that it represents a new paradigm of delivering wireless access and that it is not under the thumb of the any of the RBOCs or IXCs as most of the major wireless players are. If Metricom can successfully complete its build out, it would seem to present along with Cable TV access a strong leg of defense against telephone company ambitions.
Given the surprisingly strong PCS spectrum auctions just completed one of the worst potential pitfalls for Metricom that we can imagine is political pressure against the continuation of the unlicensed spectrum on which it depends. We note two defenses. First, it could move into unlicensed gigahertz ranges and second Dilworth has announced his intention to bid on PCS spectrum -- presumably as a backup for such a contingency.
All in all Metricom looks like a company structured in such a way as to be competitive in the new communications world rushing towards us. Decentralized, inexpensive, unregulated, plug in and play technology should enable it to be much more agile and cost effective that its huge telephony- infrastructure based competitors.
Tetherless Access - How Close to the Internet Model?Tetherless Access is the only other company that we have identified that is specifically targeting wireless access to the Internet. Located in California's Silicon valley (Fremont), it was formed in March of 1990 to develop products and services for the wireless communications market. Dewayne Hendricks (President and CEO) and Charles Brown (Chief Operating Officer) are the Joint founders of the company. Earlier this year it obtained an infusion of capital from a San Francisco-based investment firm in exchange for a minority equity share of the company. The company, operating out of offices in Palo Alto and Freemont California, has now expanded to 14 employees in preparation for bringing products to market by late this year.
According to the July 1994 TAL company profile TAL's mission is: "to help expand the benefits of the Internet to new users worldwide by delivering economical solutions on multiple platforms for high speed, wireless TCP/IP connectivity." "At the core of the business is a proprietary software technology by which a basic spread spectrum packet radio - a "virtual wire" typically used for telco by pass with a range of up to 20 miles - is transformed into a TCP/IP - compliant packet radio routing device." TAL has a partnership with a California based manufacturer [Cylink] of spread spectrum radio equipment [902-928 Mhz] to produce a packet radio router/modem called the SubSpace 2001 with a through put into the customer's computer of 64,000 bits per second.
While the price of 3,500 dollars is considerably more expensive than the Metricom radios, the speed is greater and range longer. The TAL radio can easily function on a line of sight basis of 20 miles and has reached 30 miles. The radio/router itself measures 9 by 9 by 3 inches and is intended to be used with a roof top mounted directional or omnidirectional antenna.
The TAL prospectus states that "the TAL network will consist of the following:
End nodes with TAL SubSpace units located on customers' premises and interconnected with customers' host computers and LANs via industry standard interfaces. Initially these will be fixed point sites, athough TAL's intention is to accommodate mobile devices in the future.
Base stations employing TAL technology and operated by TAL or another service provider (a TAL alliance partner). These will be established at appropriate locations within a metropolitan area to provide adequate geographic coverage to all customer sites.
Interconnections among the base stations, either using wireline or wireless technology, that ultimately link to the Internet.
Within the TAL network each node not only sends and receives its own traffic but also relays the transmissions of nearby nodes. In other words the TAL network operates as a mesh network rather than in a star/hub configuartion that is more commonly used for wireless networks. The efficiency of TAL's software allows relaying to take place while holding latency to acceptable levels within the network. This approach mnimizes transmitter power and, consequently, the interference to nodes at other locations in the network that are trying to send traffic at the same time."
"As the node density in a given area increases, the average distance between nodes decreases." Under these conditions "the SubSpace 2001 automatically adjusts its power output downward, thus decreasing the potential interference. This effectively increases the carrying capacity of the network to help accommodate the additional users. In other words the TAL approach is highly scalable."
TAL is targeting the following markets: (1) small to medium sized businesses that want a high speed internet connection at less cost that that of a telephone company based leased line connection; (2) businesses with multiple satellite sites in an urban area that are looking for TCP/IP and Internet connectivity for the satellites as well as for the "main office"; (3) large businesses that rely on TCP/IP networks and are in need of a backup to their wireline systems as part of a disaster recovery plan; (4) businesses operating in areas where the telecommunications infrastructure is poor or non existent; (5) K-12 school systems, community colleges, and libraries looking for an economical way to interconnect their schools and branches and deliver the internet to their students and the public.
In an August 1 telephone interview Dewayne Hendricks told us that between now and year's end TAL will be opening a commercial wireless network in the south bay area of San Francisco and one in the Arlington Virginia area. He also pointed out that TAL had completed a beta test for Apple Computer's Library of Tomorrow program where they linked together some of the public libraries in San Diego. In Arlington TAL has a joint venture with VITA (Volunteers in Technical Assistance) which will involve ground links for a third party low earth orbiting satellite venture in return for a share of the satellite bandwidth.
The Telluride Infozone ProjectIn cooperation with Apple Computer TAL is due to install within the next two weeks wireless connections to Telluride's local computer net. We interviewed Richard Lowenburg the Project Director.
Lowenburg told us that he came to Telluride in the late 1970s at the invitation of John Lifton and Pamela Zoline who were in involved in town planning. We have been telling the local folks to consider their communications infrastructure as being just as critrical to their future as they would sewers and utilities. In 1985 we formed the Telluride Institute which counts folk like Amory Lovins and John Naisbitt on its board and is involved in a number of issues having to do with community building.
We have been conceptualizing a project that we call the Infozone as a means of modeling at least some of what an information society would be like. About three years ago with the beginning of the information highway talk our project became real. We got some funding from the Colorado Advanced Technology Institute as the first Rural Colorado Telecommunications Program they funded. CATI initailly had the creation of a telework center in mind. We suggested that the project should fund an internet connection instead and see what grew from that.
In 1992 we got some fast track funding from them and at about the same time we were awarded an Apple Library of Tomorrow grant which placed public access terminals in around eight different locations in the community. The CATI award subsidizes our connection to the Internet (a 56 kbs POP to Colorado Supernet) and provides us with some development funds. It looks hopeful that our funding through CATI will run for another two years.
We do want to extend outside the town of Telluride itself and are very tuned into to the work of Dave Hughes and Frank Odasz in Big Sky Telegraph. We have no specific intent to model our work here after them but we certainly share common ideas and goals. Our project covers the entire local calling area of about 60 by 60 miles and serves ranching and mining communities as well as the resort of Telluride. Now the majority of our users are in Telluride and in the town of Norwood (via an Apple system in the Agricultrural Extension Office) about 37 miles from here.
We are so small and so isolated that we see ourselves as ideal for modeling an entire community network. Thus among the 1500 year round citizens of Telluride we brought our schools into this, our historical museum, our local and county government, the medical center and the arts community. About 150 people have accounts on the system. To grow further we believe that we have to improve applications and content of what we offer. Our library has one of the Apple systems of course and is well plugged into CARL and CARL Uncover.
Starting late last year various people including Steve Cisler at Apple had been indicating to Tetherless Access that they ought to be looking at what we were doing as a good testbed for TAL to explore. Dewayne came here during the winter took a look around and agreed to proceed. They plan to give us high speed wireless connectivity between each of our apple-donated systems at the library, conference center, medical center, schools, the public access radio station and so on. The wireless connected machines then function as dial up terminal server nodes in the network. They have the radios and antennas in place and are finishing the necessary software to enable the system to be turned on.
[Later the same evening Hendricks explained to us that the software is serial port drivers for the Macintoshes that will enable them to accept data at rates greater than 56 kbs. He anticipated that they are a week or less away from completion and that the network should be turned on by August 15. Norwood he added would be reached by an intermediate radio placed at a ridge top that would then beam down to Norwood on the other side.]
So What Difference Will Wireless Make?The technology to make a difference is clearly here. The infrastructure is not. And the rules by which the infrastructure will be built and technology sorted out are not yet clear.
Access to the Internet via wireless is clearly here. However it tends to be more expensive now than wireline access because the entire infrastructure is still telephone company dependent especially within the local loop.
Right now wireless access to the Internet via Metricom makes economic sense only to the person who really needs higher than 14.4 speed and the ability to log in to a base station easily from stationary locations were a physical connection to a phone line is not practical. (If such a person lived in New York City or in Chicago or any other place where charges per minute for local calls existed, this equation may be significantly different even now.)
Right now access to Metricom can make sense for a small service provider if Metricom in turn is bridging to a larger Internet service provider that allows resale - depending of course on what that service provider charges the Metricom attached customer. We think the same is generally true for Tetherless Access, although since they are not quite as far along with product and network roll outs as Metricom, we can not speak about them with as much certainty.
A year from now, if the Metricom roll out is successful, there should be Richochet Networks in many of our largest cities. If Metricom then links these cities via its own leased lines, it can then function as a wireless data network provider on a national scale. At this point, if it hired the technical talent, it could also function as an internet services provider and offer the most powerful form yet seen of data network by pass of the IXCs and RBOCs.
Its low cost modular technology is suited to the provision of internet access in a way that no one else's (except perhaps for Tetherless) is. Despite the appearance of some government actions to the contrary, smaller companies the size of Metricom are much better equipped to play in the Internet sphere than the huge telecom conglomerates that are about to invest billions in PCS spectrum license the cost of which will then have to be recouped from system users. Metricom is the only company that appears to have both low cost technology and a delivery mechanism unemcumbered by high cost licenses. For these reasons we can expect that it will be able to deliver Internet connectivity at or below current wireline costs.
[A reviewer of our draft commented: "I think that your conclusion is still too USA-centric to suit me. In fact, its too urban centric. You seem to forget that this is a big country and there are lots of potential Internet customers in rural areas. Wireless can make a real difference in those places. I don't see how either Metricom or TAL will be able to compete in the dense urban areas against services like residental ISDN. When you add in PCS and CDPD, its going to be a real tough competitive environment for us all. Lots of choices for the user and lots of opportunities to make mistakes."]
At some point, as far as the Internet is concerned, Metricom is likely to have to face a strategic decision of whether it itself becomes a service provider or more likely whether a separate company like RadioMail allies with it to function as a service provider.
Metricom currently makes no major difference to a regional network like BARRnet. In otherwords BARRnet has no particular reason to give good pricing to Metricom, because BARRnet and its customers are as dependent on the cost of leased lines from the telcos as they ever were. If Metricom were to achieve dense covereage of the Bay Area, the situation could change significantly to one where there would be a great deal of synergy between the two. For if BARRnet relied upon Metricom to supply much of its infrastructure, the price for wireless high speed internet connectivity in the speed range in which Metricom functions would almost certainly quickly decline below that of wireline connectivity. Of course how much of the intrfastructure BARRnet could rely on is subject to debate. It is unlikely that large corporations would be willing to give up their T-1 leased lines - even for multiple wireless channels operating at a fraction of that rate.
Especially as bandwidth demands continue to increase, wireless is not now and will not soon be a general solution to internet transport. But it can now fill numbers of niches in affording local loop by pass opportunities. A Metricom mesh can provide individual service in any area where there is a cluster of users - urban or rural. Metricom radios don't daisy chain well. However the Tetherless radios do. Under these conditions they may provide cost effective solutions to getting from Telco pops to rural users. Furthermore the involvement of Tetherless with Vita and the low earth orbiting satellites needs to be closely watch and applauded. Finally it would be technically possible to bridge a Tetherless Radio daisy chain via a Metricom base station into a Metricom network, if Metricom which has not sold its base station technology yet were to make the bridge available at a reasonable cost.