At IBM's annual
Corporate Technical
Recognition Event (CTRE), the company
rewards achievements that have provided exceptional value for the company or represent significant advances in science and technology. The Research Division featured strongly in this year's event, held in Naples, Florida, in late May. Three new IBM
Fellows were appointed from Research, and 11 other individuals from the division received awards. The honors recognized achievements ranging from devising a new method of data compression to designing antivirus technology. Other individuals
received IBM Corporate Patent Portfolio Awards, which recognize inventors whose patents significantly enhance the value of IBM's patent portfolio.
Bijan Davari-
A leader in
high-performance
logic technology
Bijan Davari has worked with semiconductors for most of his adult life. Born in Tehran, Iran, he received his Ph.D. from Rensselaer Polytechnic Institute, where he completed his thesis on the interface behavior of semiconductor devices. On joining IBM's Thomas J. Watson Research Center in 1984, he began working on ways to improve Complementary Metal-Oxide-Semiconductor (CMOS) technology, which provides the basis for much of today's semiconductor processing.
In 1985, Davari began the task of defining IBM's next generation of CMOS integrated circuits, which came to be called CMOS-5X. One approach to achieving the goal of integrating more transistors onto a single piece of silicon is to make the devices smaller. However, simply scaling down the dimensions of a device won't work if the voltage powering the circuits remains the same. The heat generated by the densely packed transistors at higher voltages would become excessive, and the higher voltage in itself would cause reliability problems, resulting in the malfunctioning of the devices. So as devices grow smaller, their working voltages have to be reduced.
At the time Davari started, the emerging industry standard voltage was 3.3 volts. Meanwhile, several technological advances made by Davari and his colleagues were pointing the way to a new voltage standard. Dual-polysilicon gates provided better control of current through small devices, while shallow-trench isolation offered a way to build electrical separators between transistors to allow for greater circuit densities. "These innovations removed the barriers to reducing voltages and enhanced the packing densities," explains Davari.
Beyond those barriers, the trick was to determine a new standard voltage to provide low power and adequate reliability without sacrificing performance. The answer emerged quickly. "The sweet spot for the 0.25-micron devices was 2.5 volts," Davari recalls. That voltage not only generated less heat, it also produced a device that was more reliable. Indeed, the device actually gained in performance. "We changed the perception in the industry," says Davari. "We showed that, if the geometries were scaled down properly, you could scale down the voltage and gain significant performance."
IBM introduced CMOS-5X in 1994. It serves as the basis for the PowerPC® 601+ and at least nine other microprocessors, including the microprocessors used in the latest announcement of the IBM System/390 servers.
Today, along with the advanced CMOS development, Davari spends his time on system scale integration, figuring out how to use all the transistors that can now be integrated onto a chip. "We are moving toward putting a whole system on a single chip," says Davari. "With this kind of integration, the gains in performance can be substantial. You could create tiny devices that connect you to huge networks. Also, high-end servers will be significantly affected because most of the hardware just goes onto a very few silicon chips."
Bruce Lindsay -
Making Databases
More Useful
Ever since his days as a doctoral candidate in computer science, Bruce Lindsay has been interested in the problems of large systems. Thus it was natural that, when he joined IBM in 1978, he began working at the San Jose Research Laboratory, which at the time was just completing the research on relational databases that became the foundation for IBM's SQL(TM) and DB2® database products. Lindsay sees databases as big systems that have big - and intriguing - problems associated with them. The question he always asks is, "What can you do to make them more useful?" Since users over the years have applied databases in different ways, Lindsay and his colleagues have had to come up with appropriate architectural changes to answer that question.
"Up through the late 1980s," he points out, "databases were concerned with traditional record-keeping. That is still important, but in the 1990s we've seen people wanting to use the information in their databases to run their businesses. They want to use data to profile customers or position products on the shelves; so we needed to make available wider cuts of data." Customers' needs drove the demand for distributed databases that would allow a user to find information in different places with a single query. In 1988, by which time he had moved to the Almaden Research Center, Lindsay helped define the Distributed Relational Database Architecture(TM) (DRDA®) protocols, which produced the most efficient and robust distributed database processing in the market.
His work on distributed databases revealed a significant point. "Because data is a company's most important asset, companies will always strictly control who can have access to that data," he explains. "So the problem with distributed databases is getting information from multiple sites managed by separate organizations. You have to be able to move data around while still maintaining its integrity."
The advent of nonalphanumeric types of data has also changed the design of database architecture. The question, says Lindsay, is how to describe data that takes the form of, say, a picture or a video. "Suppose you want to search for pictures of horses facing west at sunset," says Lindsay. "How do you characterize these objects?"
As the principal architect of Starburst, the first extensible database, Lindsay dreamed up and developed the concept of "database extenders." These treat multimedia data - such as images, voice and audio - as objects that are extensions of the standard relational database and can be queried using the standard structured query language (SQL). These particular database extenders are currently undergoing beta testing.
Lindsay, meanwhile, is now involved in planning Version 3 of DB2. He is also thinking ahead. "In the back of my mind," he says, "I would like to work on high-dimensional indexing, which would allow databases to perform approximate matches on complex objects such as images, text and genomes."
Ted Selker -
Making Computers
Easier To Use
"What I care about is people," says Ted Selker, with the passion of a man uttering a manifesto. This passion explains in large part why Selker has spent so much of his time trying to make computers more usable.
A native of Spokane, Washington, Selker began his career at Brown University, where he studied brain modeling in his work toward a degree in applied math. While his first two theses were on brain modeling, his Ph.D. had a more practical bent of using artificial intelligence to create adaptive interfaces. His natural drive to create things started him out designing tools, first at Brown & Sharpe, then at Stanford and Atari, where he worked with Alan Kay, before moving to the Xerox Palo Alto Research Center. He joined the IBM Thomas J. Watson Research Center in 1985, and moved to the Almaden Research Center in 1992.
Even before joining IBM, Selker had been thinking about designing a computer help program that would adapt to the needs of each user. The computer would know enough about an individual user to frame its advice at the appropriate level. Better still, the computer would change or adapt its understanding based on the user's behavior. As the user learns more, the level of advice changes so that it is always appropriate. Selker devised a system that he called COACH (for COgnitive Adaptive Computer Help). It formed the subject of his Ph.D. thesis, and became a part of IBM's OS/2® operating system in 1996.
But, as Selker points out, he is always tinkering with ways to make computers easier to use. He spearheaded development of the TrackPoint® pointing device for the ThinkPad® laptop computer. And in recent years he has become involved in designing computers that take the idea of usability one step further. One of his inventions allows a user to snap off the rear-cover panel of a ThinkPad so that the flat-panel display can be placed on an overhead projector and the display's contents can be projected onto a wall or screen. Some of these ideas revolve around the notion of carrying wallet-sized computers that could be strapped to a user's waist or carried in one's pocket. But Selker's imagination has led him to explore many other scenarios as well. "I try to visualize new paradigms for using computers," says Selker.
One result is known as Mall Walker, a personal communications assistant (PCA) that shows where its owner is in a mall or theme park, keeps track of children (as dots on a map), invites merchants to "bid" electronically on items on your shopping list as you pass their stores, and gives more information on sales and prices when you actually enter one.
But Selker's dominant theme is bringing people and computers closer at all levels. "I am interested in computers that try to build a relationship with you," he explains. "When you look at your e-mail, imagine a party. There's somebody you haven't seen in 30 years, a gaggle of colleagues, a lot of people you've never met. You have to make very precise decisions about whom you want to talk to and whom to avoid. By analyzing your past, the computer can help you make those decisions. It could teach you how to browse the Web and eventually do it for you." If even a small fraction of Selker's ideas reach fruition in products, the relationships between humans and their computers will grow increasingly more natural.
Designers of Data Sharing
Chandrasekaran Mohan and Inderpal S. Narang, both of whom have worked at the Almaden Research Center since 1981, shared a corporate award of $325,000 with four members of the Software Solutions Division in Santa Teresa. Their achievement: designing and implementing Information Management System and DB2® data sharing for System/390® Parallel Sysplex(TM).
IMS(TM)/DB and DB2 provide highly scalable database management for IBM's System/390 MVS® Parallel Sysplex, using an innovative data-sharing
architecture. This is especially important to large, high-growth customers on S/390®, the majority of whom use IMS/DB and DB2 today. This required significant innovation, including the shaping of the sysplex itself for the needs of database management, and inventions in the areas of concurrent data access, locking, data and buffer coherency, shared extended storage and performance.
Vanquishing Viruses
William C. Arnold, David M. Chess, Jeffrey O. Kephart, John F. Morar and Steve R. White, all of the Thomas J. Watson Research Center, shared a corporate award of $100,000 for their innovative contributions to the science, technology and systems design of IBM's AntiVirus family of international software products that detect and repair computer viruses. The family embodies world-class contributions to the science and technology of detecting viruses and repairing the damage they cause, as well as significant innovations in multiplatform systems design. The family operates in OS/2®, Windows 3.1, Windows 95, Windows NT, DOS and Novell Netware environments. It is also noted for extremely fast product cycles: a new release on each platform reaches the market every three months.
Communications among Applications
Richard Dievendorff of Almaden shared a Corporate Award with six members of IBM's Networking Software Division. The $275,000 award recognizes technical innovation in the architecture, design and delivery of the MQSeries(TM) product family.
Reliable application-to-application communications is critical to many of today's business processes. The MQSeries has rapidly made IBM the leader in the fast-growing market for message-oriented middleware. It provides functions for which many large customers have a critical need: the ability to communicate between applications, and to run on a variety of platforms, in a time- and location-independent manner.
In his 19 years at IBM, Dievendorff has worked on system memories, data encryption algorithms, message managing and other projects related to personal computers. He carried out much of his award-winning work during an assignment at IBM's Networking Services Division facility in Hursley, England.
Expanding the
Keyboard
John P. Karidis, a former Watson researcher now with IBM's Consumer Division, received a corporate award of $50,000 for his innovative design concept for the expandable TrackWrite (Butterfly) keyboard and his leadership in engineering development of the product. The keyboard, introduced in the IBM ThinkPad 701, represents a revolutionary improvement in compact packaging of mobile computers by providing a full-size keyboard in a subnotebook computer with a 10.4-inch display. The TrackWrite has won widespread praise in the industry. The development process, which demanded adherence to stringent restrictions on design, reliability, and manufacturability, took the concept to implementation as a product in less than two years.
Prowess in Patents
Last year, for the third consecutive year, IBM received more U.S. patents than any other company. In fact the 1995 total of 1,383 U.S. patents awarded to the company represents the single largest number of patents ever granted to any company in a single year.
Jeffrey D. Gelorme, manager of the Applied Polymer Technology Group at Watson, played a significant role in amassing that number. He notched seven patents last year, covering all aspects of electronic manufacturing, from printed circuit boards to logic levels in semiconductor chips. His personal favorite: photoresist used in the manufacture of logic elements in CMOS technology. For his patent contributions, Gelorme received a Patent Portfolio Award.
Meanwhile, Mark N. Wegman, a 21-year veteran at Watson, received a $40,000 Corporate Patent Portfolio Award for the substantial licensing value obtained from his U.S.
Patent Number 4,814,746. The patent covers an improved data compression method that significantly increases the efficiency of data transmission between, for example, host computing systems and remote terminals. It has found use in such IBM products as TERSE - a means of sending files in compressed form - and in software support for large disk storage systems. IBM has licensed the patent to numerous other manufacturers. In addition, the patent has been adopted as a key element of the CCITT V.42.bis standard for data compression for modems.
