Peer-to-Peer Traffic Dominates
Summaries of Research and Inventions from Science and Technology Journals
Using a newly designed monitoring system, researchers at Sprint Corp., in Overland Park, Kan., recently investigated how the company's data backbone is being used and found that during 2001 and 2002, peer-to-peer applications, such as Gnutella, generated up to 80 percent of the traffic on some of its links. The researchers also learned that streaming media accounted at times for a quarter of the traffic on some links but did not compare with peer-to-peer or Web traffic, which swung between 11 and 90 percent. More generally, they found that most of the links are working at less than 50 percent of capacity and that the data transmission times are dominated by the speed of light rather than by any traffic-related delays. Put succinctly, Sprint's network has plenty of backbone: you can't blame it if your voice-over-Internet Protocol (VoIP) software experiences delays.
Packet-Level Traffic Measurements from the Sprint IP Backbone , by Chuck Fraleigh, et al ., IEEE Network , November 2003, pp. 6--16.
Beijing, host of the 2008 Olympic Games, has a traffic problem. The number of passenger vehicles in use in China has been increasing at a staggering rate of 30 percent annually, pushing aside bicycles in major cities. So getting a handle on urban traffic control ahead of the 2008 deadline is a serious issue. The engineers working on the problem have already dedicated thousands of electronic devices to Beijing traffic, including TV monitors, messaging systems, intersection controllers, and a variety of others. Ideally, the end result will be a system that uses real-time information to manage traffic. Much remains to be accomplished, but the project still has a few years until the eyes of the world are upon it.
Toward Intelligent Transportation Systems for the 2008 Olympics, by F.-Y. Wang, S. Tang, Y. Sui, and X. Wang, IEEE Intelligent Systems , November 2003, pp. 8--11.
Searching technical literature effectively is something of an art form, albeit a daunting and frustrating one. Searching the biological technical literature--now growing exponentially due to automated generation of genetic data by DNA sequencers, microarrays, and other devices--poses a particular challenge. To combat this problem, scientists at the University of Oklahoma and the University of Texas have developed a novel algorithm to search large databases intelligently. To analyze how cohesive a set of objects are --such as proteins, medicines, or genes--the algorithm statistically compares the extent to which the connections between these objects in the literature deviate from random chance. This strategy quickly gives researchers a first-order estimate of how previous studies have linked the objects, potentially saving hours of analysis. The authors of the algorithm have quickly put it to practical use: implementing it in proprietary software to aid drug discovery.
Shared Relationship Analysis: Ranking Set Cohesion and Commonalities within a Literature-derived Relationship Network , by Jonathan D. Wren and Harold R. Garner, Bioinformatics , 22 January 2004, pp.191--98.
Engineers at STMicroelectronics and the French government-run laboratory CEA-LETI recently reported making a radio-frequency (RF) microelectromechanical system (MEMS) switch integrated directly onto a silicon IC, a first that could make for more versatile and compact wireless devices. The switch is used to direct signals to different radio elements, such as antennas or filters, so that a device can shift between telecommunications standards. Generally, these switches are based on gallium arsenide PIN diodes and field-effect transistor switches, which cannot be integrated onto a silicon chip. But building the switch directly onto the IC reduces parasitic impedances that can diminish or distort the signals going through off-chip switches. The engineers were able to pull off two key achievements: combining the fabrication of an electromechanical switch with standard silicon IC processes and keeping the switch well packaged and protected. The new switch can handle 3 watts of power at 2.4 gigahertz.
An Above IC MEMS RF Switch , by D. Saias, et al. , IEEE Journal of Solid State Circuits, December 2003, pp. 2318--24.