How much is it worth to let us do it

[David Farber <farber () linc cis upenn edu>]

become increasingly crucial.  Cellular systems that spurn
Qualcomm today may find themselves in a quagmire of TDMA
microcells tomorrow.  Together, all the gains from CDMA bring
about a tenfold increase over current analog capacity.  In
wireless telephony above all, wide and weak will prevail.
     
     Like any obsolescent scheme challenged by a real innovation
- and like minicomputers and mainframes challenged by the PC -
TDMA is being sharply improved by its proponents.  The inheritors
of the Linkabit TDMA patents at Hughes and International Mobile
Machines Corp.  (IMMC) have introduced extended TDMA, claiming a
19-fold advance over current analog capacity.  Showing a
conventional cellular outlook, however, E-TDMA fatally adopts the
idea of increasing capacity by lowering speech quality.  This
moves in exactly the wrong direction.  PCN will not triumph
through compromises based on a scarce-spectrum mentality.  PCN
will multiply bandwidth to make the acoustics of digital cellular
even better than the acoustics of wire-line phones, just as the
acoustics of digital CDs far excel the acoustics of analog
records.
     
     Riding the microcosmic gains of digital signal processing,
CDMA inherently offers greater room for improvement than TDMA
does.  Bringing the computer revolution to cellular telephony,
CDMA at its essence replaces frequency shuffling with digital
intelligence.  Supplanting the multiple radios of TDMA - each
with a fixed frequency - are digital-signal-processing chips that
find a particular message across a wide spectrum swath captured
by one broadband radio.
     
     With the advance in digital electronics, the advantage of
CDMA continually increases.  As the most compute-intensive
system, CDMA gains most from the onrushing increases in the cost-
effectiveness of semiconductor electronics.  Qualcomm recently
announced that it has reduced all the digital signal processing
for CDMA into one application-specific chip.
     
     For all the indispensable advances of CDMA, however,
Qualcomm cannot prevail alone.  It brilliantly executes the move
to digital codes, but proprietary mainframe computer networks are
digital, too.  As presently conceived, CDMA still aspires to be a
cellular standard using the same mainframe architecture of mobile
telephone switching offices that now serve the analog cellular
system.  In itself the Qualcomm solution does little to move
cellular toward the ever cheaper, smaller and more open
architectures that now dominate network computing and will shape
PCN.






Hearing Feathers Crash Amid Heavy Metal


     
     Consummating the PCN revolution - with its millions of
microcells around the globe and its myriad digital devices and
frequencies - will require a fundamental breakthrough in cellular
radio technology.  In the new Steinbrecher minicell introduced
early this month at the Cellular Telephone Industry Association
show, that breakthrough is at hand.  The first true PC server for
PCN, this small box ultimately costing a few thousand dollars
will both replace and far outperform a 1,000-square-foot base
station costing more than a million dollars.
     
     Once again, in an entrepreneurial economy, crucial
innovations come as an utter surprise to all the experts in the
field.  Donald Steinbrecher began in the Radio Astronomy Lab at
MIT in the 1960s and early 1970s, creating receivers that could
resolve a random cosmic ray among a mass of electromagnetic
noise.  This required radios with huge dynamic range - radios
that could hear a feather drop at a heavy metal rock concert.  He
and his students solved this intractable problem by creating
unique high-performance receivers and frequency "Mixers."  These
could process huge spans of spectrum with immense variations of
power and translate them without loss into intermediate
frequencies.  Then, computer systems convert the signals from
analog to digital and analyze them with digital signal
processors.
     
     Moving out to begin his own company in 1973, Steinbrecher
and his colleagues made several inventions in the fields of radar
and digital signal analysis.  At first, most of their customers
were national security contractors in the intelligence field.
For example, Steinbrecher supplied the radios for the ROTHR
(remote over the horizon radar) systems that became famous for
their role in the war against airborne drug traffic.  Then in
1986, the company was asked if its equipment could work in the
cellular band.
     
     After cosmic rays and battlefield radar, the cellular band
was easy.  When he saw that the digital signal processors at the
heart of his systems were dropping in price tenfold every two
years, Steinbrecher knew that his esoteric radios could become a
consumer product.
     
     Translated to cellular, this technology opens entire new
frontiers for wireless telephony.  Rather than tuning into one
fixed frequency as current cellular radios do, Steinbrecher's
cells can use a high-dynamic-range digital radio to down-convert
and digitize the entire cellular band.  TDMA, CDMA, near or far,
analog cellular, video, voice or data, in any combination, it
makes no difference to the Steinbrecher system.  His minicell
converts them all at once to a digital bit stream.  The DSPs take
over from there, sorting out the TDMA and CDMA signals from the
analog signals and reducing each to digital voice.  To the extent
the Steinbrecher system prevails, it would end the need for
hybrid phones and make possible a phased shift to PCN or a
variety of other digital services.
     
     Hoping to use Qualcomm chipsets and other technology,
Steinbrecher could facilitate the acceptance of CDMA.  For CDMA,
the minicell provides a new, far cheaper radio front end that
offers further relief to the near-far problem and is open to the
diverse codes and fast-moving technologies of PCN.  For the
current cellular architecture, however, Steinbrecher offers only
creative destruction, doing for large base stations what the
integrated circuit did for racks of vacuum tubes in old telephone
switches.
     
     In essence, the new minicell replaces a rigid structure of
giant analog mainframes with a system of wireless local area
networks.  Reconciling a variety of codes and technologies, the
Steinbrecher devices resemble the smart hubs and routers from
SynOptics Communications and Cisco Systems that are transforming
the world of wired computer networks.
     
     Best of all, at a time when the computer industry is
preparing a massive invasion of the air, these wide and weak
radios can handle voice, data and even video at the same time.
Further, by cheaply accommodating a move from scores of large
base stations to scores of thousands of minicells per city - on
poles, down alleys or in elevator shafts - the system fulfills
the promise of the computer revolution as a spectrum multiplier.
Since each new minicell can use all the frequencies currently
used by a large cell site, the multiplication of cells achieves a
similar multiplication of bandwidth.
     
     Finally, the Steinbrecher receivers can accommodate the
coming move into higher frequencies.  Banishing once and for all
the concept of spectrum scarcity, these high-dynamic-range
receivers can already handle frequencies up to the "W band" of 90
gigahertz and more.






Boundless Bandwidth


     
     The future of wireless communications is boundless
bandwidth, accomplished through the Shannon strategy of wide and
weak signals, moving to ever smaller cells with lower power at
higher frequencies.  The PCN systems made possible by Qualcomm
and Steinbrecher apply this approach chiefly to voice and data.
Recent announcements by Bossard and Hovnanian extend the concept
to television video as well.  Last December, they disclosed that
their company, Cellular Vision, was already wirelessly delivering
49 cable television channels to 350 homes near Brighton Beach,
Long Island, in the 28-gigahertz band.  They declared a plan to
soon sign up some 5,000 new customers a month all over New York.
     
     Among engineers in cellular and cable firms, Cellular Vision
evokes the same responses of incredulity and denial familiar at
Qualcomm and Steinbrecher.  Like them, Bossard is resolutely on
the right side of the Shannon and Shockley divide.  In answer to
the multitude of qualms and objections and demurrals, all three
companies cite the huge benefits of more bandwidth.  Qualcomm can
assign some 416 times as much bandwidth to each call as a current
cellular or TDMA system.  Steinbrecher's minicell receivers can
process 4,160 times as much bandwidth as an analog cell site or
TDMA radio.
     
     Hovnanian achieves some 300 times the bandwidth of a
broadcast TV station and some three times the bandwidth of even a
typical cable head end.  For Hovnanian's so-called multipoint
local distribution system, the FCC has allocated a total of two
gigahertz between 27.5 and 29.5 gigahertz - one gigahertz for TV
and one gigahertz for experimental data and phone service.  This
large swath of spectrum allows Cellular Vision to substitute
bandwidth for power.  Using FM rather than the AM system of
cable, Cellular Vision gains the same kind of increased fidelity
familiar in FM radio.  Assigning 20 megahertz to each channel -
three times the six megahertz of an analog system - Cellular
Vision proves the potency of wide and weak by getting 20 decibels
- some 10 times - more signal quality.  These extra decibels come
in handy in the rain.
     
     With a radius of three miles, Cellular Vision cells are
about 100 times smaller than telephone cells.  Transmitting only
10 milliwatts per channel over a three-mile radius, the system
gets far better signal-to-noise ratios than the three-watt radios
of cellular phones or the multikilowatt systems of AM radio or
television broadcasts.  The millimeter wavelengths at 28
gigahertz allow narrowband high-gain antennas that lock onto the
right signal and isolate it from neighboring cells.  At 28
gigahertz, small antennas command the performance of much larger
ones (for example, a six-inch antenna at 28 gigahertz is
equivalent to a three-foot antenna at 4 gigahertz or a 300-foot
antenna at broadcast television frequencies).
     
     In Brighton Beach the receiving antennas, using a fixed-
phased-array technology, are just four inches square, and the
transmitting antennas deliver 49 channels from a one-inch
omnidirectional device on a box the size of a suitcase.  Between
cells, these transmitters can send programming and other
information through a conventional point-to-point microwave link.






Singing in the Rain


     
     So what happens in the rain? Well, it seems that Cellular
Vision does better than conventional cable.  When you have small
cells in geodesic low-power wireless networks using the full
computational resources of modern microchips, you have plenty of
extra decibels in your signal-to-noise budget to endure the most
violent storms.  Indeed, the 350 Brighton Beach customers of
Cellular Vision received continuous service during the November
1992 near hurricane in New York, which brought floods that
interrupted many cable networks for hours.  One competitive
advantage of Cellular Vision over cable seems to be less
vulnerability to water.
     
     Moving television radically toward the regime of wide and
weak, Bossard and the Hovnanians have changed the dimensions of
the air.  However, they cannot escape the usual burdens of the
innovator.  Any drastic innovation must be some 10 times as good
as what it replaces.  Otherwise, the installed base, engineering
momentum and customer loyalty of the incumbents will prevail
against it.
     
     Cellular Vision faces a wired cable system with some $18
billion in installed base.  Already deploying fiber at a fast
pace, cable companies plan to move within the next year toward
digital compression schemes that increase capacity or resolution
by a factor of between six and 10 (depending on the character of
the programming).  That means some 500 digital channels or more.
TCI, the leading cable company, has ordered some one million
cable converter and decompression boxes from General Instruments'
Jerrold subsidiary for delivery late in 1993.  In the U.S. cable
industry, Hovnanian faces an aggressively moving target.  Most
cable experts doubt he can make much of a dent.
     
     This view may be shortsighted.  Clearly, Cellular Vision -
and its likely imitators - can compete in the many areas with
incompetent cable systems, in areas yet unreached by cable or in
new projects launched by developers such as the Hovnanians.  In
the rest of the world, cable systems are rare.  Cellular Vision
is finding rich opportunities abroad, from Latvia to New Zealand.
Most of all, as time passes, Cellular Vision might find itself
increasingly well positioned for a world of untethered digital
devices.
     
     Such a cellular system could be adapted to mobile telephone
or computer services.  With a bit-error rate of one in 10
billion, it could theoretically transmit computer data without
error correction.  With one gigahertz of bandwidth, the system
could function easily as a backbone for PCN applications,
collecting calls from handsets operating at lower frequencies and
passing them on to telephone or cellular central offices or to
intelligent network facilities of the local phone companies.
     
     The future local loop will combine telephone, teleputer and
digital video services, together with speech recognition and
other complex features, in patterns that will differ from
neighborhood to neighborhood.  Easily customizable from cell to
cell, a system like Bossard's might well offer powerful
advantages.
     
     In an era of bandwidth abundance, the Negroponte switch -
with voice pushed to the air and video onto wires - may well give
way to this division between fibersphere and atmosphere.  With
the fibersphere offering virtually unlimited bandwidth for fixed
communications over long distances, the local loop will be the
bottleneck, thronged with millions of wireless devices.  Under
these conditions, a move to high-frequency cellular systems is
imperative to carry the increasing floods of digital video
overflowing from the fibersphere.
     
     In any case, led by Qualcomm, Steinbrecher and Cellular
Vision, a new generation of companies is emerging to challenge
the assumptions and structures of the existing information
economy.  All these companies are recent startups, with
innovations entirely unexpected by international standards
bodies, university experts and government officials.  They are
the fruit of an entrepreneurial America, guided by the
marketplace into the microcosm and telecosm.






Why Imitate European Failures?


     
     Meanwhile, the European and Japanese experiences with
government-guided strategies should give pause to proponents of
similar policies here.  Thirty years of expensive industrial
policy targeting computers has left the Europeans with no
significant computer firms at all.  The Japanese have done
better, but even they have been losing market share across the
board to the U.S.
     
     In the converging crescendos of advance in digital wireless
telephony and computing, progress is surging far beyond all the
regulatory maps and guidebooks of previous years.  If the entire
capacity of the 28-gigahertz band, renewed every three miles, is
open to telephony and video, bandwidth will be scarcely more
limiting in wireless than it is in glass.
     
     In this emerging world of boundless bandwidth, companies
will prevail only by transcending the folklore of scarcity and
embracing the full promise of the digital dawn.  In an era of
accelerating transition, the rule of success will be self-
cannibalization.  Wire-line phone companies are not truly
profitable today; their reported earnings all spring from slow
depreciation of installed plant and equipment that are fast
becoming worthless.  As George Calhoun of IMMC demonstrates in
his superb new book, Wireless Access and the Local Telephone
Network (Artech, 1992), new digital wireless connections are
already less than one-third the cost of installing wire-line
phones.  For the RBOCs, aggressively attacking their own
obsolescent enterprises is their only hope of prosperity.
     
     As Joseph Schlosser of Coopers & Lybrand observes, self-
cannibalization will not appear to be in the financial interests
of the established firms; it will not prove out in net-present-
value terms.  There will be no studies to guarantee its success.
Executives will have to earn their pay by going with their gut.
As semiconductor and computer companies have already learned,
phone and cable companies will discover that self-cannibalization
is the only way to succeed in this era - the only way to stop
others from capturing the heart of your business.
     
     This is the lesson of the last decade.  When Craig McCaw
sold his cable properties and plunged into cellular telephony and
$2 billion of Michael Milken's junk bond debt, there was no way
to prove him right.  Today AT&T is preparing to launch him as a
rival to Bill Gates as the nation's richest man.  Yet McCaw
cannot rest on his laurels; the hour of the cannibal is at hand.
     
     In theory, the transition should not be difficult for this
resourceful and ingenious entrepreneur, who has long been a
leading prophet of ubiquitous wireless phones and computers - his
predicted personal digital assistant, "Charles."  But a company
that has paid billions for its 25-megahertz national swath of
long and strong frequencies faces especially acute dilemmas in
moving toward a regime of wide and weak.  As a man - and company
- that has made such transitions before, McCaw is favored by
history and by AT&T.  As a giant pillar of the new establishment,
though, McCaw may find it as difficult to shift gears as did the
computer establishment before him.  The stakes are even higher.
The next decade will see the emergence of fortunes in ever-
changing transmutations of PCN, digital video, multimedia and
wireless computers that dwarf the yields of cable and cellular.
The window of opportunity opens wide and weak.