The Cutting Edge: Computing / Technology / Innovation : Software Predicts Radio Wave Action
Wireless computer networks hold the promise of eliminating expensive wiring and cabling in office complexes and granting the flexibility to move equipment around without rewiring. But transmitting interference-free radio signals within buildings is a complex engineering problem. A new software package developed by engineers at Georgia Tech Research Institute to predict how radio waves transmit and reflect inside structures could help solve this problem.
The software, called a cell engineering tool, was developed through an applied research program sponsored by Hitachi Telecom USA Inc. It helps make educated judgments about where to locate the wireless base stations within a building. The ability to quickly choose the best locations should minimize the number of stations necessary and thus reduce the system’s cost. To come up with their experimental measurements, researchers measured and characterized radio signal propagation in high-rise office structures, several Georgia Tech buildings, a few warehouses and the Inforum building in downtown Atlanta. The information was used to develop a baseline set of information that was then used to predict how a given signaling system would work.
While the software will not work for every building, researchers hope to refine it to work with a large enough percentage of possible cases to make it a valuable tool.
Eric Barnhart, a Georgia Tech research engineer, and Les Pickering, another senior research engineer, are considering starting a company to commercialize the technology.
From Hubcaps to Lasers: The shiny, chrome-plated bumpers and hubcaps that adorned classic cars such as the Chevrolet Bel-Aire, Ford Fairlane and porthole Thunderbird may ultimately pave the way for a new generation of semiconductors so small they can be measured by layers of atoms.
The process of electrodeposition, which uses an electrical current to deposit a layer of metal onto another substance, has been around for a long time. But though it’s good for chrome bumpers, it doesn’t have the degree of control needed to form an electronic-grade material.
Now John Stickney, an electrochemist at the University of Georgia, has developed a technique that may make it possible to electrodeposit materials for a wide variety of electronic devices, including the lasers in compact disc players or detectors for infrared instruments, heat-seeking missiles and night vision goggles.
For the last five years, Stickney has been experimenting with ways to alternately electrodeposit one-atom-thick layers of two or more elements in a 1-to-1 ratio to form a compound--rather than using the crystal structures employed in traditional electrodeposition. His recently patented technique, Electrochemical Atomic Layer Epitaxy, is more environmentally friendly than other deposition methods because it doesn’t produce a lot of toxic waste gases.
And theoretically, the technique should cost less since it uses smaller amounts of chemicals. One compound Stickney has experimented with--cadmium telluride--has the potential to produce very efficient photovoltaic cells because the compound has the ability to absorb light very effectively.
Better Computer Brains: Computers were once known as “electronic brains,” but it soon became obvious that the human brain works very differently than a computer. Neural networks have attempted to electronically mimic how brain cells function, but commercializing them has been slow work. For one thing, the real neural networks of the brain use a massive point-to-point 3-D interconnection scheme that is impossible to realize on a two-dimensional chip.
The traditional alternative has been a time-shared “bus,” or data channel, that switched between the various points to be connected. But normal digital busing schemes destroy the analog characteristics of a signal--especially timing, which is critical for the brain to detect motion. Imagine a train moving toward you. The brain perceives this motion because the visual (analog) signals arrive over time. Should the image of the train arrive digitally in one burst, the train would appear to be motionless.
Now Applied Neurodynamics, based in Encinitas, Calif., has announced that it has managed to fold the three-dimensional topology of the brain into a two-dimensional representation appropriate for semiconductor chips, without destroying the continuous nature of time.
A digital-signal processor attaches time stamps to each pulse as it emerges from an analog neuron. The time stamp is broadcast onto the bus along with its destination code. Circuitry at the destination neurons pick out only the signals intended for them and synchronize them in time-stamp order. A board-based prototype proving the concept has been designed with researchers at Caltech and Britain’s Medical Research Council.
