News Briefs

Pages: pp. 23-25

TECHNIQUE USES VIRUSES TO MAKE COMPUTING COMPONENTS

A startup company is combining organic and inorganic substances in a new technique for making computing and electronic components.

Cambrios Technologies is developing components by combining inorganic materials currently used in solid-state electronics with genetically engineered variants of the M13 bacteriophage—a virus that attacks the E.coli bacteria, a strain of which is commonly associated with food poisoning, explained CEO Mike Knapp.

Figure    Cambrios Technologies has developed a technique for combining organic and inorganic substances to make less-expensive, higher-performing computing and electronic components. Cambrios genetically engineers variants of a virus, which produce different proteins. The company's process determines which of the proteins will bind to a semiconductor material. Cambrios continues the process until it determines the candidate proteins that best bind to the material. In production, the material binds to the protein, which the virus-generation process has shaped into the desired component's form.

According to Knapp, biochemistry is a new and useful tool in electronics production that lets manufacturers assemble higher-performance, lower-cost nanoscale structures. The process is less costly because, unlike traditional silicon-based manufacturing, it works at room temperature and also doesn't use hazardous or toxic chemicals. In addition, the viruses and bacteria used are inexpensive.

By working effectively at nanoscale, biochemical-based manufacturing can make components to precise specifications, thereby yielding higher performance.

Nanotechnology journalist Howard Lovy, editor of the NanoBot Web site ( http://nanobot.blogspot.com/), said the semiconductor industry generally would prefer to use organic materials in its production process because they can be more easily manipulated and modified at the molecular level, giving scientists the ability to tailor their properties as needed.

As Knapp explained, Cambrios uses standard genetic-engineering techniques to create variants of the M13 virus, utilizing E.coli as a host for the process. Each M13 variant produces a different protein. Cambrios wanted to generate proteins because some have affinities for binding to particular metals and semiconductor materials, Knapp noted.

"You create lots of random changes," Knapp said, "and select the ones that have the desired properties, like the ability to bind to inorganic materials of interest." Cambrios has experimented with materials including metals used in electronic components such as gold, platinum, silver, and cobalt, as well as substances typically used in semiconductor manufacturing such as silicon and germanium.

The metal or semiconductor material binds to the protein, which has been shaped into the proper form for the desired component by the virus-generation process.

Cambrios also examined whether these interactions could develop nano-wires by creating a protein that binds to and thus takes the shape of a tubular virus. "If you heat it up, the organic materials burn off, leaving a structure in the size and shape of the virus," Knapp explained.

He said the computing-related applications for which Cambrios' techniques could be utilized include nano- wires and thin films of material used in making chips and other electronic components.

Cambrios plans to work first on developing field-effect transistors for field-emission displays, a new type of flat-panel display.

RESEARCHERS TO MOVE SPEECH RECOGNITION TO HARDWARE

A university researcher is leading a team developing a chip that provides speech recognition, a task previously handled only in software.

The work, led by Carnegie Mellon University professor Rob A. Rutenbar, could markedly change speech recognition, mobile communications, and even security and emergency-response applications.

Currently, recognizing arbitrary speech—as opposed to words or brief phrases preprogrammed into a system—requires complex, computationally intensive software run on a powerful CPU. "It requires lots of MHz, Mbytes, and memory," explained Rutenbar.

Such power-hungry CPUs won't work in energy-strapped, battery-powered mobile devices such as cellular phones or machines used in the field by emergency and other workers.

Thus, Rutenbar noted, cellular phones currently have only limited speech-recognition capabilities. However, an advanced voice interface would be desirable in such devices, which are too small for useful keyboards or mice.

Rutenbar said placing speech recognition on an efficient chip dedicated to the task, rather than placing it in complex software run by the CPU, would enable advanced mobile-device interfaces and generally improve handheld voice capabilities by "two to three orders of magnitude."

One of the Carnegie Mellon project's key goals is to create a new speech-chip architecture that performs the task 100 to 1,000 times more efficiently than software run on a CPU. A dedicated speech chip, using advanced semiconductor technology, would run slower than a CPU and perform computations in parallel, thereby consuming less energy. This would make the chip suitable for battery-operated handheld devices.

Putting voice capabilities on hardware is one of the technology's Holy Grails, said Judith Markowitz, president of J. Markowitz Consultants, a speech-industry consultancy. However, Rutenbar said, figuring out how to do this efficiently in hardware without using too much energy has been a challenge.

The Carnegie Mellon researchers are also trying to make the speech-recognition process itself more efficient in three areas: the digital signal processing used in handling the speech input, the matching of the input against the representations of sounds in the system's database, and the eventual identification of words and sentences.

Rutenbar said the technology could be used in many ways, such as in cellular phones, hands-free communications devices for emergency workers, and voice-controlled universal remote controls.

The research team hasn't decided whether to license the technology or develop products themselves, he noted.

NEW SOFTWARE USES INSTANT MESSAGING TO PLAY GAMES

Game designer Jules Urbach has developed software that he says could change the world of both online gaming and instant messaging (IM).

Urbach's software, Otoy, is an engine that lets participants play games over IM. He said this could not only change online gaming but also begin moving IM beyond chat into other, more lucrative areas, a long-time goal of many messaging proponents.

IM has worked with very simple games such as tic-tac-toe in the past. However, Urbach wanted to integrate his complex, 3D, multiplayer games into IM to encourage users to play the games with people on their buddy lists, thereby spreading the games widely and quickly and also encouraging word-of-mouth marketing.

To allow Otoy to work with instant messaging, Urbach explained, "We have millions of lines of code hooking into IM."

Otoy also uses a proprietary, lossless image-compression system that significantly reduces the size of games and makes them easier to play over IM, he noted.

Urbach said he is designing small-footprint games for Otoy because they're even quicker and easier to run over IM applications. Also, he noted, "In general, the smaller the game, the greater the number of people we will see taking the time to play it."

Otoy could be used in gaming communities or Web sites that include games among their offerings. Otoy will generate revenue, Urbach explained, by attracting advertising to participating Web sites and enabling companies to sell virtual components or objects for games, such as a faster engine for an automobile in a racing game.

While Urbach predicts success for Otoy's new approach, Nitesh Patel, analyst for Strategy Analytics, a market research firm, said, "Gaming over instant messaging is not something that has any great potential. I don't think carriers are going to push IM because it cannibalizes their revenues." Carriers would rather encourage the use of text messaging because it is more profitable, he explained.

In addition, Patel said, other technologies used for gaming, such as Java, are better established. Moreover, he added, selling inexpensive game components one at a time may be too complicated to succeed widely.

Thus, he said, "I struggle to see how IM is going to succeed as a gaming platform." However, he noted, it's possible the approach may develop a niche following.

In 1999, Urbach and two partners founded Groove Alliance, one of the first gaming companies that created 3D products exclusively for online use. He also designed Hell Cab, one of the first CD-ROM games.

Sewing Machines Go Digital

In the 200 years since they were invented, sewing machines gradually moved from manual to electric operations. Now, sewing machine manufacturers are rapidly adding computer technology, which is beginning to play an important role in the industry.

One of Bernina's new sewing machines, for example, uses a 206-MHz Intel StrongArm processor. The embroidery component alone has 4 Mbytes of memory. The machines also have a CD-ROM drive and a USB port to connect to a computer.

Bernina's Artista 200E sewing machine is computerized, complete with Intel StrongArm processor, Windows CE operating system, CD drive, USB port, and touch screen.

The high-end versions use the Windows CE operating system. However, each machine's features can be supplemented by the capabilities of a computer to which it connects. Because the machines run Windows, which many users already know, the learning curve often is not great.

The devices run software programs, many for digitizing images such as photographs or artwork that will serve as designs on items. Users can transfer the digitized designs to the sewing machine via portable memory or a cable connection from a PC or laptop. The machine can then render the designs as a series of stitches in an embroidered piece of work.

Most computerized sewing machines have touch screens, displaying adjustable features such as stitch width and length, thread tension, and speed. Some machines automatically set variables for users based on the type of fabric or stitch, a feature that can be overriden.

Conventional sewing machines are often large, slow, and difficult to use. They can chew up fabric and frequently take hours to complete even relatively simple designs. Their computerized counterparts, on the other hand, make it easier and quicker to do even complex embroidery and stitch patterns, as well as quilting, which used to require specialized machines run by sewing professionals. Even on routine jobs, computerized machines are much faster than traditional ones.

The new equipment also automates manual operations that were routine but time-consuming on traditional machines, such as needle threading, thread cutting, and holding fabric in place between stitches.

Computerizing formerly complex tasks frees home sewers to be more creative in garment construction, noted Hans Herzog, Bernina's technical education manager.

Russell Scott, owner of and teacher at the Sewing Arts Center in Southern California, said that computerizing sewing machines has led to a renaissance in home sewing. "Sewing machines made sewing easier," Scott explained. "Computerization has bumped that up tenfold."

The new generation of computerized machines cost between about $600 and$7,500, noted Herzog.