YOU鈥橵E HEARD the tape and played the CD, now listen to the credit card. Paul Simon鈥檚 Graceland album has never sounded better. But that鈥檚 not the point, says Howard Singer. He鈥檚 one of the developers of the slim, solid-state memory card storing Simon鈥檚 tunes. 鈥淲hich would you rather jog with?鈥� he asks music industry executives as he slips the card neatly into a small player no bigger than a pager: 鈥淎 disc player that skips when you nudge it, a tape player that bounces around on your belt, or a solidstate player the size and weight of a pager that never misses a beat?鈥�
There鈥檚 no choice, of course, as there never is with the best sales pitches, and Singer hardly needs to confirm his conviction that solid-state memory cards will replace tapes and CDs: 鈥淚t is not a question of if but when moving media become extinct,鈥� he tells the disc and tape makers. They are shocked, as you might expect, and yet their surprise has as much to do with the messenger as it has with his message. For standing before them is a scientist from Bell Laboratories, the research wing of the American telecommunications giant AT&T and not a fellow executive.
But Singer鈥檚 sales pitch typifies the new face of Bell Labs, the organisation that can boast seven Nobel prizewinners, has averaged a patent a day over its 70-year life, and gave the world the transistor, the laser, the negative-feedback amplifier, the solar cell, the light-emitting diode, the digital telephone exchange, the communications satellite, the cellphone and talking pictures. Despite this record of achievement, Bell Labs can no longer afford to sit back and see what turns up from its research, as it has done in cosy isolation for most of its life at three sites in New Jersey 鈥� Murray Hill, Holmdel and Middletown (see 鈥淥rigins of a creative giant鈥�).
Advertisement
Since AT&T was stripped of its commercial monopoly of the nation鈥檚 telephone service in 1984, the corporation has had to work hard to compete with new rivals. And its prize exhibit remains its research division, whose expertise is now plugged into the demands of the real world as never before.
Typical of the new approach are AT&T鈥檚 plans for the music industry, using Singer鈥檚 solid-state recording and pager-sized player. While most electronics companies that promise a glamorous future for the technology rely on the availability of cheaper chips with larger memories, Singer bases his predictions on new ways of compressing digital data so that less memory is needed in the first place.
AT&T鈥檚 compression technology, known as Perceptual Audio Coding (PAC), was not developed with solidstate recording in mind, but as a national standard for digital audio broadcasting. PAC splits the audio signal into narrow frequency bands, more than a thousand in all, but digitises only those frequencies that we can hear. Musicam, a rival system developed in Europe, works in the same way but from a base of only a few hundred bands. As a result, PAC provides a more accurate analysis.
鈥淭he people who developed PAC were hi-fi enthusiasts,鈥� says Schuyler Quackenbush of the Signal Processing Research Department at Bell Labs, who is currently trying to persuade broadcasters in the US to chose PAC rather than Musicam as their industry standard for DA8. 鈥淭hey were given a free hand. That鈥檚 the nature of Bell Labs, despite the new economic reality.鈥�
AT&T鈥檚 eagerness to establish its credentials with a wider audience has led it to refurbish some old buildings at Murray Hill to make way for what it calls the Human-Centered Engineering Laboratory. Despite the grandiose title, the new laboratory has a very down-to-earth brief: to get the names of 鈥淏ell鈥� and 鈥淎T&T鈥� to mean something to the general public. 鈥淧eople now buy a computer because it says 鈥業ntel inside鈥�,鈥� says its director, Greg Blonder. 鈥淲e want people to buy telecommunications equipment because it says 鈥楢T&T inside鈥�.鈥�
Like Singer, Blonder knows the publicity value of attention-grabbing gadgets and rarely goes anywhere on business these days without his Dick Tracy wristwatch telephone. At exhibitions, Blonder is regularly seen talking into the cuffs of his shirt or showing off some other gadget, such as his 鈥渃hameleon鈥� telephone that changes colour when it rings or to match its surroundings. Although these prototype telephones currently work only with a base station, extrovert executives can expect the next models to be fully fledged cellphones. By 1997, for instance, Bell Labs plans to have shrunk the essential circuitry for the Dick Tracy telephone to the size of a watch case and to have designed it to run on just 2.5 volts so that the batteries can squeeze in, too.
Away from the slick sales patter and wacky gadgets, Bell Labs is helping AT&T to develop its core business 鈥� known as POTS to insiders, or plain old telephone service 鈥� at an astonishing pace. Ten years ago, modems could drive only around 300 bits per second through the twisted pair of copper wires that still links the telephone network into most homes. Now they can manage around 30 000 bps and, by the end of the century, the rate could be a million bps, says Victor Lawrence, who leads the Advanced Multimedia Communications Department at Bell Labs.
The problem for the early modems was that they could recognise only 1 bit of information for every digital on-off pulse transmitted. Circuits capable of recognising 8 or even 9 different levels for each pulse have since been developed, and this has increased modem transmission rates to 28 800 bps. While the same technology could be tweaked to lift the rate to 50 000 bps, which would enable domestic telephone lines to transmit data as efficiently as a modern digital link, Bell Labs has set its sights higher to meet consumer demands. Lawrence forecasts that telephone subscribers will soon want home links capable of transmitting 1 million bps: 鈥淭he driving force is the merger of communications and entertainment. The pace of growth is phenomenal. Delivering megabit streams into the home will be commonplace in the next five years.鈥�
This growth will have significant repercussions. As soon as homes start to consume megabit streams, the demand on the national and international trunk networks will outstrip the capacity of the fibre-optic cable that carries the signals, the amplifiers that boost them on long runs and the switches that route them to the dialled destinations.
Engineers at Bell Labs have been looking at these issues since 1991, when AT&T laid an experimental network of fibre-optic cable across the US to develop new communications equipment. The idea was to exploit a system for packing more information into a datastream, known as asynchronous transfer mode. ATM allows streams of data 鈥� audio, video, text and graphics 鈥� to intermingle during transmission, and so make the most of the available capacity.
But while such tricks enable digital code to be transmitted at great speed as light beams along optical fibres, they can do little to maintain the flow through the electronic amplifiers. These boosters are required every 70 kilometres or so to drive the signals on their way, but they become bottlenecks as the light pulses are converted into electronic pulses, amplified and then converted back again into light pulses. 鈥淎t a gigabit a second, an electronic amplifier cannot cope,鈥� says Jay Simpson of the Optical Fiber Research Department.
Bell Labs took this problem seriously enough to expand a small team already developing an optical amplifier from just four or five in the mid-1980s to several hundred. Their efforts seem to be paying off: last year, AT&T claimed to have laid the world鈥檚 first all-optical undersea link, which stretches 2000 kilometres along the seabed from St Thomas in the Virgin Islands in the Caribbean to West Palm Beach in Florida.
Simpson makes optical amplification sound simple. Just mix the weakening pulsed signal with a powerful steady beam, and then guide the light through a 20-metre loop of optical fibre doped with erbium, the rare earth element that can transfer energy from one light beam to another. The principle of optical amplification was known thirty years ago, says Simpson, but that was before AT&T had invented solid-state lasers to deliver the energising beam and before it had found a way of utilising erbium鈥檚 potential.
Early light pumps, he recalls, were 鈥渁 Frankenstein array of enormous flash lamps鈥�. The boosters now sitting on his desk at Murray Hill are the size of a portable radio, and look absurdly simple. A few electrical components are needed to control the laser diode and that鈥檚 it, so there is little to go wrong, says Simpson. Sadly life is never that simple. The erbium loop adds random noise to the signal, which becomes intolerable at data speeds above 5 Gbps. But Bell Labs thinks it has the answer to this problem, too.
Linn Mollenauer works at Holmdel, about half an hour鈥檚 drive away, and he is investigating solitons. These are solitary waves that have the unusual property of being able to propagate without spreading out. Mollenauer began this research in 1991, and was one of the first people to demonstrate how the property might be used to reduce unwanted background noise in telephone transmissions (鈥淭he secrets of everlasting life鈥�, New 杏吧原创, 15 April).
The proving ground sits under his laboratory bench. Several coils of optical fibre, each the size of a small car wheel, together provide a transmission line 5000 kilometres long. Pieces of string appear to tie erbium amplifiers to the coils. By running a signal twice round the loop, Mollenauer鈥檚 bench mimics a fibre-optic cable that stretches 10 000 kilometres across the bed of the Pacific Ocean. Using solitons, he says, the loops can already handle a datastream running at 20 Gbps and will be ready to go online by the time the world needs the capacity.
Bell Labs is still learning how to balance the new commercial necessities with the desire to maintain something of its early tradition of pure scientific investigation. 鈥淓ven five years ago, that鈥檚 five years after divestiture, researchers in these labs didn鈥檛 know what AT&T made,鈥� says Blonder. He compares the atmosphere in those days to the one that 15th-century artists in Italy enjoyed under the Medicis: 鈥淪omeone, somewhere was paying scientists to work on whatever they liked. Science for science sake, unconstrained by relevance to AT&T 鈥� that kind of ivory-tower research has now stopped.
Origins of a creative giant
AMERICAN Telephone and Telegraph Company, AT&T, dates from March 1885 when the American Bell Telephone Company, set up ten years before by Alexander Graham Bell to capitalise on his invention, created the subsidiary to provide long-distance calls. For the next hundred years, AT&T expanded and ran the Bell System, the nation鈥檚 telecommunications network.
In 1925, riding on the back of this profitable monopoly, it opened the Bell Telephone Laboratories at Murray Hill in New Jersey. Because it was a legally sanctioned monopoly with regulated profits, the corporation could write off the institute鈥檚 costs as a business expense, and so it gave the fledgling research centre the freedom to investigate any topic however loosely connected with communications. By the 1960s, the researchers had added so many new wings to the original laboratories that they had run out of space at Murray Hill, and they spilled over into the nearby sites of Holmdel and Middletown.
That era ended on 1 January 1984, ten years after the US Department of Justice had filed an antitrust suit calling for the break-up, or 鈥渄ivestiture鈥�, of AT&T鈥檚 national telecommunications monopoly. The corporation鈥檚 seven regional services, or Baby Bells, became independent local networks, and other companies such as MCI and Sprint began to compete for the lucrative business of connecting long-distance calls.
While AT&T managed to hang on to Bell Labs, it lost around 3000 scientists and engineers who left to join the Bell Communications Research Group, or Bellcore, which the Baby Bells set up. Witthin a couplee of years, the number of staff at Bell Labs had fallen by about a quarter, from around 26 000 to under 20 000.
But now the dust seems to have settled, and AT&T is again investing heavily in research. Annual spending is $3.5 billion, equivalent to roughly five per cent of turnover, compared with $2.5 billion, or four per cent of turnover, at the time of divestiture. Furthermore, the total number of staff at the various research sites has risen to 25 000.