Association for Computing Machinery
Welcome to the August 1, 2011 edition of ACM TechNews, providing timely information for IT professionals three times a week.

Also, please download our new ACM TechNews iPhone App from the iTunes Store by clicking here and our new ACM TechNews iPad App by clicking here.


Face-ID Tools Pose New Risk
Wall Street Journal (08/01/11) Julia Angwin

Carnegie Mellon University (CMU) researchers are studying how facial-recognition tools can be detrimental to privacy. In a recent test, the researchers were able to identify about 33 percent of the people they tested, only using a snapshot and facial-recognition technology from Google. In addition, CMU professor Alessandro Acquisti found that he could use information available on Facebook to correctly predict the first five digits of a person's Social Security number about 27 percent of the time, demonstrating the potentially intrusive potential of facial-recognition technology when used with publicly available personal data. Acquisti says the study shows how Facebook is becoming a de facto identity-verification service. As part of the CMU study, 93 student volunteered to be photographed with a webcam. The pictures were uploaded to a cloud computer and put into a database of 261,262 publicly available photos taken from CMU students' Facebook profiles. The researchers were able to find 10 possible matching photos in the Facebook database with more than 30 percent accuracy. The research "suggests that the identity of about one-third of subjects walking by the campus building may be inferred in a few seconds combining social network data, cloud computing, and an inexpensive webcam," Acquisti says.
View Full Article - May Require Paid Subscription | Return to Headlines

Progress Hits Snag: Tiny Chips Use Outsize Power
New York Times (07/31/11) John Markoff

A research team recently presented a paper at the International Symposium on Computer Architecture that highlights the problem of developing smaller, more densely packed transistors to continue the rapid development of electronic devices. The paper, co-authored by Microsoft Research's Doug Burger, who chairs ACM SIGARCH, says that soon the most advanced chips will have so many transistors that it will be impractical to simultaneously power them all, producing a situation in which some of the transistors are left without power while the others are working. The problem has the potential to disprove Moore's Law, which states that the number of transistors that can fit on a chip will double about every two years. However, some researchers say the problem will be solved. "The good news is that the old designs are really inefficient, leaving lots of room for innovation," says NVIDIA chief scientist William J. Dally. For example, Intel recently developed a way to vary the power consumed by different parts of a processor, enabling slower and less power-consuming transistors and faster-switching cores to work simultaneously. Intel says that future chips will have different cores for specific problems, only some of which will require high power.

Steve Perlman's White Paper Explains 'Impossible' Wireless Tech
VentureBeat (07/28/11) Dean Takahashi

Inventor Steve Perlman's distributed input distributed output (DIDO) technology enables each wireless network user to use the full data capacity of shared spectrum concurrently with other users by removing interference between users sharing the same spectrum. Perlman explains in a white paper how DIDO operates, and the paper notes that DIDO is far less costly than a conventional commercial wireless deployment and features higher capacity and performance. It also can harness existing consumer Internet infrastructure and indoor wireless access points--thus supporting a virtually unlimited number of users. DIDO involves the collision of radio signals at users' computers, and the sum of the signals at each computer's location results in a clean modulated waveform conveying the data intended for that specific computer. Each computer then demodulates the signal waveform and plays the transmission. DIDO also employs advanced math rather than switching to enable users to move between adjacent DIDO networks without needing handoffs. The system's intelligence is centrally located in the data center, which communicates to all users simultaneously through all of the access points at once. DIDO access points also can transmit much longer distances than Wi-Fi access points or cell towers.

Nanowire Electronics That Can Be Shaped to Fit Any Surface and Attach to Any Material Developed at Stanford
Stanford Report (CA) (07/28/11) Louis Bergeron

Stanford University researchers have developed a method of attaching nanowire electronics to the surface of almost any object, regardless of its shape or what material it is made of. The researchers say the technique could lead to wearable electronics, flexible computer displays, high-efficiency solar cells, and ultrasensitive biosensors. The key to the technique is coating the surface of the silicon wafer with a thin layer of nickel before fabricating the electronic circuitry. Nickel and silicon are both hydrophilic, which means that when they are exposed to water after fabrication the water penetrates the two materials, detaching the nickel and the electronics from the wafer. "The detachment process can be done at room temperature in water and only takes a few seconds," says Stanford professor Xiaolin Zheng. After detachment, the silicon wafers are clean and ready to reuse, which could greatly reduce manufacturing costs. The researchers also used an ultrathin layer of an extremely flexible polymer to serve as an insulator and provide mechanical support for the electronics. The short length of the nanowires used to fabricate the circuitry is what allows the devices to bend with the flexible substrate, according to the researchers.

EU Project Examines the Scalability and Performance of I/O Subsystems in Multicore Platforms
Institute of Computer Science (07/27/11) Angelos Bilas

European Union researchers are working on IOLanes, a project designed to improve input/output (I/O) performance in hardware that utilizes multicore architectures. IOLanes researchers are examining how multicore central processing units can be used to design new persistent I/O architectures. IOLanes has designed I/O subsystems that will result in 10 times fewer storage nodes in data centers and 10 times greater workload capacity for the same number of storage nodes. IOLanes researchers are targeting the performance and scalability issues of the I/O stack on multicore architectures; addressing I/O performance and dynamic resource management issues in virtualized, single-host environments; and examining the on-loading and off-loading tradeoffs for advanced functions in modern storage systems. IOLanes breaks down the I/O stack into four layers, including application and middleware, virtual machine, host operating system, and embedded storage controller. The IOLanes project will demonstrate the benefits of the new design using real-world data center-type applications.

Will Li-Fi Be the New Wi-Fi?
New Scientist (07/28/11) Jamie Condliffe

Visible light communication (VLC), which uses rapid pulses of light to wirelessly transmit information, could be an alternative to conventional Wi-Fi. The University of Edinburgh's Harald Haas says the technology is based on new high-brightness light-emitting diodes (LEDs). If the LED is on, a digital "1" is transmitted, and if the LED is off, a digital "0" is transmitted. The LEDs can be switched on and off very rapidly, making them useful for transmitting data. Researchers at the universities of Oxford and Edinburgh are using parallel data transmission with arrays of LEDs, where each LED transmits a different data stream. Other researchers are using different colored LEDs to alter the light's frequency, with each frequency encoding a different data channel. Heinrich Hertz Institute researchers have reached data rates of more than 500 Mbps using a standard white-light LED. Since VLC uses light instead of radio-frequency signals, it could be safely used in aircraft, built into medical devices, and underwater. "There are around 14 billion light bulbs worldwide, they just need to be replaced with LED ones that transmit data," Haas says. "We reckon VLC is a factor of 10 cheaper than Wi-Fi."

Protecting Networks Is Just a Game
EurekAlert (07/27/11)

A defensive strategy for computer networks based on game theory is more effective than previous methods, says Iona College information technologist Heechang Shin, who developed an anti-hacking tool that plays a game of reality versus forecast. Called defensive forecasting, the tool wins when reality matches its forecast, and then sends out an alert to block an attempt to attack the computer network. The tool works on real-time data flowing in and out of the network, rather than analyzing logs, and detects intrusions as they are happening. Shin's game theory model continuously trains the tool so that it can recognize the patterns of typical network attacks. To measure the effectiveness of the tool, Shin compared it using the semi-synthetic dataset generated from a raw TCP/IP dump data by simulating a typical U.S. Air Force local-area network to a network intrusion system based on a support vector machine (SVM), which is one of the best classification methods for detection. During testing, the tool was as good or better than one based on SVM for detecting network intrusion while adding the benefits of real-time detection.

Carnegie Mellon Develops iPhone App That Predicts When Bus Will Arrive
Carnegie Mellon News (PA) (07/27/11) Byron Spice

Carnegie Mellon University researchers in the Rehabilitation Engineering Research Center on Accessible Public Transportation (RERC-APT) have developed Tiramisu, an iPhone application that uses crowdsourcing to enable riders to know when the next bus will arrive. Tiramisu displays the nearest stops, a list of buses and light rail vehicles that are scheduled to arrive, the estimated arrival times, and the level of fullness, as indicated by the riders already on the vehicle. The system also helps riders with disabilities by indicating if there is room on the bus for a wheelchair. In testing, the RERC-APT team found that even a small number of riders can provide useful information about the buses, and the information becomes more useful as more riders access the system. Tiramisu is one of the initial projects developed as part of the Traffic21 initiative created by Carnegie Mellon and the Hillman Foundation. "Crowdsourcing makes it possible for riders to provide real-time updates about how the bus system is actually functioning on any given day by simply sharing information with each other," says Traffic21 director Rick Stafford.

Designing a Better Crystal Ball
Wake Forest University (07/27/11) Cheryl Walker

Researchers at the U.S. Intelligence Advanced Research Projects Activity (IARPA) and Wake Forest University are developing the Aggregative Contingent Estimation Systems (ACES), a model for crowdsourcing predictions that combines the collective knowledge of many individuals to produce more accurate conclusions. ACES relies on citizens to improve the accuracy of forecasting methods for intelligence agencies. "The idea is to combine individual judgments from a lot of people who all know a little to provide a tremendous amount of information," says Wake Forest professor Eric Stone. The researchers want individuals to contribute their knowledge in fields such as politics, the military, economics, science and technology, and social affairs, to create a more powerful prediction engine. ACES provides opportunities for anonymous sharing and deliberating before making important decisions. "Our project is designed to find out what people know, have them share this knowledge with others, and ask them to make a prediction based on what they and others know," says ACES principal investigator Dirk Warnaar.

From Detonation to Diapers: Los Alamos Computer Codes at Core of Advanced Manufacturing Tools
Los Alamos National Laboratory News (07/27/11) James E. Rickman

Los Alamos National Laboratory (LANL) researchers have developed computational tools to help make manufacturing processes more reliable. The tools are part of the federal government's advanced manufacturing initiative, which aims to make U.S. companies more competitive and create new jobs. Proctor & Gamble (P&G) has used the tools to model the flow, transport, and interaction of fluids and particles to help design a more efficient diaper manufacturing process. The tools were developed for LANL's Computational Fluid Dynamics Library, software designed to solve problems related to the dynamic behavior of materials. The tools also resulted in Reliability Technology, a comprehensive system that helped P&G reduce interruptions to production lines, which saved the company billions of dollars. "We are pleased to see that the problem-solving methodologies developed at Los Alamos National Laboratory to help ensure the reliability of the nation's nuclear weapons stockpile are being used to help ensure the reliability and competitiveness of American manufacturing," says LANL's Terry Wallace.

A Software That Teaches You How to Read
The Hindu (India) (07/25/11)

New reading software uses computer-generated voices and speech recognition to computerize a proven learning method known as repeated reading. Waikato University computer science students Ashley Steel, Luke Bjerring, and Andreister are developing the software, called BookieMonster, which is designed to act as a reading coach for children. The software reads text aloud, progressively highlighting it on-screen in time with the voice, similar to karaoke. After listening to the text a few times, children can read it back to the computer via a microphone. BookieMonster recognizes their speech and provides the same progressive text highlighting of what they received while being read to. The team has a working prototype and are setting up trials in local schools in New Zealand. They eventually hope to distribute BookieMonster to developing nations that have low literacy rates.

Microsoft Paper Proposes Using 'Cloud' Servers to Heat Homes (07/26/11) Bob Yirka

Microsoft researchers recently published a paper that proposes saving energy by installing servers used for cloud computing into homes and businesses. The researchers say the excess heat produced by the servers could be used to heat homes, which would save the hosts and cloud computing companies as much as $324 a year per server. The researchers envision having data furnaces (DFs) in the basements of suburban homes all across the United States. The servers could heat the homes, provide hot water, and be used to dry clothes, all with existing broadband connections. Larger systems with more central-processing units could be installed in commercial buildings, which would provide more computing power for the cloud company and a reduced heating bill for the host building. Such a system could result in increased storage and computing power for cloud applications without an increase in electrical demand because the electricity used to run the DFs would be offset by the electricity saved in home heating. The system also could result in faster access times for customers because the servers would be located nearby.

Abstract News © Copyright 2011 INFORMATION, INC.
Powered by Information, Inc.

To submit feedback about ACM TechNews, contact: [email protected]
Current ACM Members: Unsubscribe/Change your email subscription by logging in at myACM.
Non-Members: Unsubscribe