New Algorithm Breathes New Life into Old Browsers

 

Researchers at the University of California, San Diego, have developed Approximate Link State (XL), a new algorithm that makes routers more efficient by automatically sending network updates only to the routers that need them. In a typical corporate environment, routers flood the network with updates that every router receives. For older routers, the vast amounts of updates they receive can bring them to a halt and slow down the network while they process the updates and recalculate their path tables. To combat the problem, IT staff can manually isolate router groups and configure them to receive and process only those network updates within their group. The XL algorithm eliminates the need for manual configuration of these groups by withholding or forwarding updates it determines are necessary to its group. The trade-off is that routers don’t have precise information on the actual state of the network. The XL algorithm works with Intermediate System-to-Intermediate System and Open Shortest Path First link-state routing, allowing for interoperability with existing router protocols. For use, router manufacturers will have to integrate the algorithm into their software. The researchers presented their research at SIGCOMM 2008 in August; the paper is available here. (Computerworld)

F# Brings the Functional Programming

 

In a community technology preview (CTP) released in August, Microsoft is showcasing the F# language’s improved integration, libraries, and scripting. F# is a functional programming language for the .NET platform that also supports object-oriented programming. Functional programming languages emphasize the application of functions rather than changes in state like in imperative programming. The CTP includes better integration with Visual Studio and allows for improved large-scale software development. Additionally, a new feature called Units of Measure extends F#’s inference and strong typing to floating-point data. Along with the CTP, Microsoft has launched the F# Developer Center (http://msdn.com/fsharp) to provide developers with F# resources and user communities. The CTP is available for download here. (Microsoft)

Solid-State Drives Offer Faster Data Access but Raise Security Concerns

 

Because they offer low access times and latency, solid-state drives (SSDs) are becoming more popular, especially with laptop users. However, SSDs present their own special security concerns, including physical hacks. Some SSDs use NAND flash chips, which don’t have security hooks to keep the chips from being removed from their enclosures. Attackers could remove the chips, read the data with a flash chip programmer, and reassemble the data with data-recovery software. To slow down hackers, drive makers could integrate encryption keys inside SSD controller devices. Although attackers can hack encryption keys, it will add a barrier and slow down the attackers. (Computerworld)

IBM’s Project Quicksilver sets data transfer record for flash memory

 

IBM has announced that its Project Quicksilver—a research project that combines solid state flash technology and IBM’s virtualization technology—has delivered more than 1 million input/output operations per second with a response time under one millisecond. That’s roughly two and a half times faster than the fastest disk storage available today. At this speed, businesses could  complete their traditional workloads 2–3 times faster. Additionally, the technology requires only 55 percent of the power and cooling of current systems and one-fifth the space. Project Quicksilver is composed of research teams from the IBM Hursley Lab in England and the IBM Almaden Research Center in California. (ZD Net UK)

Computers learning to sort data like humans

 

Researchers at the Massachusetts Institute of Technology have developed a model that helps computers see patterns in data as humans do. The model is based on an algorithm that considers a range of data structures—including trees, linear orders, rings, dominance hierarchies, clusters, and so on—and apply them? to the data. Josh Tenenbaum, a researcher on the project, says, “Instead of looking for a particular kind of structure, we came up with a broader algorithm that is able to look for all of these structures and weigh them against each other.” The human brain does this daily, often unconsciously; children display this skill at a very young age, such as when learning that words—dog, for example—fit into several overlapping categories. Scientists could use the model to analyze huge amount of data and also to help research on how the human brain discovers patterns. (Science Daily)

Comcast will cap data usage in October

 

Starting on 1 October, Comcast, the largest provider of cable-based Internet access in the US,  will cap data usage at 250 Gbytes per month for residential broadband subscribers. Users who exceed the monthly limit will receive a call from Comcast warning them about their excessive use. Users who exceed the data limit twice in a six-month period risk having their access suspended for a year. Comcast says the limits will affect less than 1 percent of its subscribers. The 250 Gbyte limit is roughly equivalent to sending 50 million emails, downloading 62,500 songs, or downloading 125 movies, the company said. Comcast made the announcement on its Web site as a change to its Acceptable Use Policy. (Beta News)

IBM technology makes faster CPUs with reduced power consumption

 

IBM has invented technology that reduces the size of static random access memory (SRAM) cells in memory chips, which could let processors perform faster while using less power. The technology shrinks the cells to 22 nanometers and lets IBM add more functions, such as animation or 3D graphics capabilities, to processors. IBM conducted the SRAM research at the University of Albany’s College of Nanoscale Science and Engineering with several partners, including Advanced Micro Devices and Toshiba. (Computerworld)

reCaptcha captures hard to read words for book digitization

 

In a paper published online, researchers at Carnegie Mellon University detail the results of their reCaptcha program. The program uses captcha technology to help digitize old books and newspapers. The researchers report that the program is more than 99 accurate. During the program’s first year, 1.2 billion reCaptchas were solved and more than 440 million words have been correctly deciphered. The program uses optical character recognition (OCR) to scan text; when the OCR technology can’t recognize words or phrases, it creates images of the indecipherable text. These images are then used as captchas on Web sites to prevent spam or automated email registrations. The reCaptchas system compares each image to several others to determine if the deciphered words are correct. Carnegie Mellon’s researchers estimate that more than 100 million reCaptchas are solved each day and 4 million words are transcribed. (Science Daily)

Gag order stays in effect in subway hack case

 

A federal judge has refused to lift a restraining order against three MIT students in a case involving the Massachusetts Bay Transportation Authority (MBTA) and the students’  presentation showing a hack of the MBTA’s e-ticketing system. The presentation was scheduled for the Defcon convention in Las Vegas, but was scrapped when the MBTA sued to stop the presentation, claiming the students had not given it enough time or information to assess the vulnerabilities. Though the presentation was cancelled, slides from the demonstration were on a CD that Defcon attendees received and are now publicly available. The order will remain in effect until August 19 when a hearing is scheduled. (Computerworld)

WiMax helps scientists monitor volcanoes

 

Researchers from the WiMax Extension to Isolated Research Data (WEIRD) project have developed a volcano monitoring system that harnesses the power and flexibility of WiMax. The system uses the Diameter protocol, which identifies and prioritizes data from volcanic monitors so high-priority data isn’t blocked by less important network traffic. The WEIRD researchers combined WiMax with software they developed for next-generation networks so that monitoring signals will be exchanged from end to end regardless of the underlying network. For applications that aren’t designed to run on NGNs, the researchers created adaptors they call WEIRD agents or WEIRD application programming interfaces. In addition, they developed software that lets researchers easily make end-to-end connections without the need for specialized training. Using remote cameras, the researchers can also zoom in on trouble spots and receive data from them. WEIRD researchers will use the system to monitor activity from Mount Vesuvius in Italy and volcanoes in Iceland. (Science Daily)

Showing 3,651 - 3,660 of 4,468 results.
Items per Page 10
of 447