Andrew Chraplyvy had planned on springing a big surprise on the engineers and scientists assembled at an optics conference last February. He and his colleagues at Lucent Technologies’ Bell Laboratories and at&t; Research arrived ready to demonstrate a technology for transmitting 1 trillion bits of computer data per second--two and a half times faster than had ever been done before and enough to carry 12 million telephone conversations or 20 video signals over a single optical fiber.
As it turned out, when it came to surprises Chraplyvy was also on the receiving end. Two other teams, one from Fujitsu and another from ntt Corporation, had also arrived ready to break the terabit barrier. Then in September, a fourth group, from nec Corporation, outdid them all by sending a whopping 2.6 trillion bits.
Each group used a different grab bag of techniques to overcome the basic limitation: whereas an optical fiber can easily accommodate a terabit per second, the electronics commonly available for driving a laser operate at only one-fiftieth that speed. Taking the brute force approach, Fujitsu and nec broke the terabit barrier by superimposing signals from many differently colored lasers on one line. Since the at&t-Lucent; team wasn’t able to scrounge up enough spare lasers in their lab, they doubled the capacity of 25 lasers by splitting each beam into two beams of different polarizations--so that the light waves in one beam oscillated in a direction perpendicular to those in the other. Once they encoded two different signals onto each beam, they recombined them. The ntt team, on the other hand, used different colors, along with a technique for coding more than one signal at a time.
It will take another five to ten years to bring such speed to the marketplace, not least because there’s no demand for it. I don’t think anybody even knows how to use a terabit just now, says Chraplyvy.
