Timely Topics for IT Professionals
About ACM TechNews
ACM TechNews is published every week on Monday, Wednesday, and Friday.
ACM TechNews is intended as an objective news digest for busy IT Professionals. Views expressed are not necessarily those of either HP or ACM.
To send comments, please write to firstname.lastname@example.org.
Volume 4, Issue 440: Monday, December 30, 2002
- "Could Diamond Chips Supplant Silicon?"
IDG News Service (12/27/02); Miyake, Kuriko
Under the aegis of the New Energy and Industrial Technology Development Organization, the Japanese government has earmarked $6 million for fiscal year 2003 for research into diamond semiconductors. Japanese electronics companies already focusing on the development of diamond chips, such as Kobe Steel and Sumitomo Electric Industries, are expected to join the effort. The advantages of diamond-based semiconductors include higher voltage and temperature resistance than their silicon counterparts. Diamond chips can remain operational up to 1,000 degrees Celsius, while silicon chips fail when temperature exceeds 150 degrees Celsius. Likewise, the threshold for diamond chips' voltage resistance is about 200 volts, while silicon chips can only resist about 20 volts. Such advantages translate into faster speed and higher frequency capability for diamond chips, as well as the durability to operate in systems such as vehicle engines; the higher voltage resistance also allows power electronics to shrink. In addition, diamond-based flat-panel display electrodes boast higher electron output, while the use of diamond electrodes can extend the lifespan of devices. However, Hideyo Okushi of Japan's National Institute of Advanced Industrial Science and Technology notes that it will take at least two decades for diamond chips to supplant silicon because of several technical challenges that have yet to be met. Artificial diamonds cost a lot more than silicon, while the flow of electricity through diamond is uneven; engineers are trying to induce impurities to solve the second problem.
- "Hired Hackers Expose Flaws"
Investor's Business Daily (12/26/02) P. A5; Krey, Michael
Just about any computer system can be broken into given unlimited time and resources, says Fred Rica, the leader of a 130-member team for PricewaterhouseCoopers that performs threat and vulnerability assessments for companies. He compares the hacking threat to Germany's Autobahn, where accidents are few and there's no speed limit; however, when accidents do occur, "it's often a big one." Despite the risk, Rica says Internet users should use the Web with confidence, since the odds are very low that a person's credit card number will be stolen, for example. Still, he says, "We need more elegant security software. Security systems today are somewhat kludgey." He says Internet-accessible applications are often not built securely, while many of the recent problems have been caused by software bugs such as Nimba and Code Red. Today's security problems are complex, Rica says, and compounded by the threat of cyberterror, but he says the hacker population has been held in check in part by an increasing willingness to track down and prosecute hackers. Still, he cautions that corporate security efforts "have stayed about the same the last 10 to 12 years...[while] their networks have become much more complex."
- "Removable Hard Disk Group to Show New Device"
InfoWorld.com (12/26/02); Miyake, Kuriko
A prototype 1.8-inch Information Versatile Disk for Removable usage (iVDR) hard disk will be unveiled at next month's Consumer Electronics Show (CES), the first time the new removable hard disk system will be displayed outside Japan. The technology uses either a parallel or serial ATA interface to connect to consumer electronics devices, computers, audio players, and possibly car navigation systems. The parallel interface can transfer data at 100 Mbps, while the newer serial ATA interface features 150 Mbps throughput, consumes less power, and offers simpler cabling. The iVDR consortium, which now includes 28 members including hard drive manufacturers Seagate Technology and Maxtor, will also showcase a 2.5-inch iVDR drive at CES. The consortium was organized to standardize iVDR technology, promote its use, and license the technology to other vendors. Toshiaki Hioki of Sanyo Electric, a iVDR member, says the swappable removable hard disk system allows consumers to add more hard disk space as needed in an economical way. Hioki says, "I hope we can attract many computer peripheral makers at CES, so that the iVDR system will start spreading and be used in personal computers first." The consortium also plans to work with the movie industry to develop copyright control technology for the drives by March.
Click Here to View Full Article
- "2002 Marked by Sophisticated Attacks"
eSecurity Planet (12/26/02); Gaudin, Sharon
A report from F-Secure finds that the number of computer virus outbreaks this year was lower than that of last year, but the sophistication of the attacks appears to have increased. A spokesman for the security company says that the rate of detecting new viruses has more or less remained the same, while old viruses accounted for most 2002 cases. According to F-Secure, the biggest virus nuisances of 2002 were Klez and Bugbear: The former email worm has been circulating since October 2001, while the latter proliferated all over the world just a few days after its discovery this past September. F-Secure also names Slapper as the most widespread virus to target Linux systems this year, and reports an increase in activity from Asian virus creators. Symantec senior director of research Stephen Trilling warns that future trends will include more elaborate worms that can propagate in new ways, such as via Instant Messaging. Meanwhile, George Bakos of Dartmouth College's Institute for Security Technology Studies expects hybrid worms to become even more widespread, and able to exploit multiple security holes, perhaps on multiple operating systems. He anticipates additional threats from polymorphic worms that can mask their presence, as well as worms that can establish communications links with their controllers from within contaminated systems.
- "2002: The Year in Technology"
New Scientist Online (12/25/02); Knight, Will
The Year 2002 saw more efforts on the part of the music and movie industries to control the distribution and use of their content. New copy-protection technologies angered users, and even pioneering digital storage companies such as Philips, which argued that CDs with copy-protection undermined the format with intentional flaws. Hollywood and its music industry cohorts also tried to pass legislation allowing them to hack into file-trading networks and connected computers to keep them from sharing content. On the quantum computing front, the first commercial encryption device was released by a Swiss company and the U.K. defense ministry demonstrated quantum encryption keys could be sent through air, transmitting them 23 kilometers between hilltops. Australian researchers developed the first quantum-based processing device and the first quantum calculation was performed in Europe's Austria. Also on the small-tech front, IBM researchers were able to push the boundaries of what carbon nanotubes can do by applying the age-old punch-card computation method. Japan pulled far ahead in the supercomputing race by building the Earth Simulator, which can process more than 35 trillion "floating point" calculations in one second. More Chinese going online put that country in second-place behind the United States in terms of Internet usage, but the Chinese government also sought to restrict access by adding search engine Google to its list of blocked sites, though compromised access is now allowed.
- "Hi-Tech Ghosts of Christmas Future"
BBC News (12/25/02); Wakefield, Jane
In 2050, the Christmas meal will probably include synthetic turkey put together using molecular raw materials. Christmas will be different in many other ways as well, according to British Telecom futurologist Ian Pearson, due to advances in biology, nanotechnology, and other fields. Families gathering for the Christmas will all be able to interact even though some members are far away, through the use of virtual conferencing technology. Likely, gifts will be very different as well, with genetically engineered animals and animated Barbie dolls that express emotions. Pearson also expects that clothes and make-up will be imbued with intelligence so they change according to people's moods. Another ideal gift might be a dream machine that caters to people's imagination using thought recognition and contact lens displays. In any case, no one would have to make a mistake buying a gift for another person because of mind-reading technology, says Pearson.
- "10 Gig Ethernet: Speed Demon"
Computerworld Online (12/23/02); Hamblen, Matt
The10 Gigabit Ethernet technology, which the Institute of Electrical and Electronics Engineers (IEEE) approved in June, has a lot of potential: It could provide high-bandwidth remote data center replication over greater distances than single-mode fiber; it could supplant Gigabit Ethernet in server clusters and offer an alternative to InfiniBand through its use as a low-latency, high-speed backplane for blade servers; it could substitute for Fibre Channel in storage-area networks; and it could more easily set up links with telecoms' wide-area network services thanks to the frame format it shares with Synchronous Optical Network. The federal government and university research centers have adopted 10 Gigabit Ethernet technology for grid-computing projects, while Gartner analyst Mark Fabbi predicts that the standard may eventually be embedded into corporate data centers. Most organizations have been reluctant to adopt 10 Gigabit Ethernet because of high implementation prices--a typical 10 Gigabit port costs an average of $50,000--but analysts expect a 30 percent to 50 percent reduction in cost within the next 12 months. In fact, Foundry Networks' Chandra Kopparapu believes the cost per port will be $5,000 to $6,000 by 2006. The IEEE is considering supporting 10 Gigabit Ethernet over copper, but Gigabit Ethernet Alliance President Richard Brand says the standard's coverage is unlikely to exceed 25 meters, and adds that there could be issues with electromagnetic interference. The increasing popularity of desktop-accessible 10 Gigabit Ethernet will cause an increase in 10 Gigabit deployments for corporate LANs, according to Yankee Group analyst Zeus Kerravala.
Click Here to View Full Article
- "InfiniBand: What's Next?"
eWeek Online (12/23/02); Burt, Jeffrey
Vendors and customers alike are taking a wait-and-see approach to InfiniBand, although smaller companies have begun releasing some components needed for an InfiniBand environment. Rather than migrating data center architectures, companies are considering gradual upgrades using PCI-X and 10 Gigabit Ethernet as a way to improve performance. Meanwhile, InfiniBand suffered a serious setback when major supporters Microsoft and Intel announced they would halt new development of the technology. However, a handful of smaller firms, including Voltaire, InfiniCon, and InfiniSwitch, already have products in the pipeline that will enable enterprises to make the switch. A few large vendors, such as IBM, Sun, and Dell, remain committed to rolling out InfiniBand-fitted hardware next year or in 2004. IBM says its entire eServer line will come InfiniBand-ready in 2003, and the company will build an Intel-based eServer xSeries machine with InfiniBand host channel adapter, switch, and fabric management. Dell and Sun plan to make their blade servers compliant with InfiniBand technology, and Sun will build InfiniBand support into its storage, switch, and server product lines. Los Alamos National Laboratory plans to test InfiniBand technology beginning in January, deploying 128 servers in its supercomputing infrastructure. Team leader Mike Boorman says InfiniBand is relatively cheap and connects with just about seven microseconds of latency, compared to 10 to 20 microseconds with Ethernet. Hewlett-Packard remains reticent, as chief technology officer Karl Walker says his company is deciding whether to back Ethernet-based products or InfiniBand.
- "Mind Games"
IEEE Spectrum (12/02); Cass, Stephen
Video games are taking advantage of artificial intelligence in order to become smarter and more entertaining; GameAI.com editor Steven M. Woodcock explains that such qualities will help game companies be more competitive as 3D graphics and features that add realism to games become widespread. Through AI, video-game characters can learn more and interact more smoothly with human players. One of the more popular forms of game AI is finite-state intelligence, in which a non-player character (NPC) follows a limited number of behaviors or states determined by the game designer in response to specific triggers: For instance, the NPC enters a Charge state when a player comes into view, or switches to a Flee state when its health is low. Fuzzy logic or random weighting can be embedded into the NPC's decision-making process to give its actions a sense of unpredictability. Another game AI technology, scripting, uses a high-level language to define the control of NPCs and game events; it can either be a fixed and linear series of actions that often set up cut scenes, or a more intricate program that can deploy other kinds of AI, such as finite-state machines. Unlike AIs being developed academically, game AIs can cheat, often using qualities of the virtual worlds they operate in--worlds that feature built-in clues and commands that academic AIs purposefully lack. Advanced AI is particularly important for open-ended and strategy games, while massively multiplayer role-playing games and first-person shooters also stand to benefit. Relatively new to the market is AI middleware, which is supposed to provide game developers with a customizable AI program.
- "Robust Speech Recognition at KAIST"
Speech Technology (12/02) Vol. 7, No. 6, P. 40; Lee, Soo-Young; Kim, Yoon
The Brain Neuroinformatics Research Program at the Korea Advanced Institute of Science and Technology (KAIST) is a multidisciplinary effort to understand how biological brains process information and to develop intelligent machines whose functions are modeled after such mechanisms. The program's current focus is developing noise-robust speech recognition technology using the human auditory central nerve system as a guide. Researchers have diagrammed the system into a schematic in which signals from both ears are combined at superior olivery complexes (SOCs) and are channeled into auditory cortexes via the inferior colliculus; signal processing between the SOC and auditory cortex is only vaguely understood at this point. An artificial auditory system must use three integrated paths--an object path, an attention path, and a spatial path: The object path is made up of sub-processes--nonlinear feature extraction, time-frequency masking, time adaptation, and complex feature functions; the attention path consists of bottom-up (BU) and top-down (TD) attention, and the model currently being worked out combines both mechanisms; and the spatial path is comprised of sound localization and noise reduction with binaural processing. For this last process, models calculate interaural time delay, which is used to localize sounds and reduce noise, while deconvolution and separation algorithms are added to account for multipath reverberation and multiple sound sources. A field programmable gate array (FGPA) version of the binaural processing model has been deployed, and an application-specific integrated circuit (ASIC) version will debut soon. The various models are thought to be ready for commercial use, since they exhibit considerable auditory recognition performance gains in noisy environments.
- "The Biggest Hole in the Net"
Newsweek Online (01/06/03); Brill, Steven
Newsweek columnist Steven Brill writes that civil libertarians have been so appalled at the idea of a standard national ID card that they do not even discuss it, even though debate is critical to the development of a system that both shields civil liberties and upholds security. Furthermore, polls have shown that most Americans are not opposed to the idea of an ID card standardized by the government. Brill argues that discussing a national ID system now is essential, given the likelihood that one day terrorist suicide bombers will strike in the United States, forcing a halt in commerce and other daily routines while massive searches by low-wage guards are conducted. Although preventing the first attack is impossible, Brill agrees with Stephen Flynn of the Council on Foreign Relations that having a complete security system in place will assure Americans by its very presence, and its promise that it will thwart the next assault; the only other option is to stop all activities following an attack until the system is finished. There has been virtually no progress made in devising a system that can pre-screen people arriving at thousands of access points where large groups of people gather. The cost of setting up such a system would be outrageous, and the delays it would cause would be intolerable; Brill writes that in such situations, a national ID card may be the only viable option. The cards would come with a biometric identifier, while card applicants would be checked against continuously updated government criminal and watch-list databases. The development of such an ID system should come out of a government push, but the private sector could issue the cards.
To read more about ACM's actions in regards to national ID cards, visit http://www.acm.org/usacm.