Dr. Dobb's Journal June 2002
Students from 67 countries competed in this year's Association for Computing Machinery (ACM) International Collegiate Programming Contest, but top honors went to a team from China's Shanghai JiaoTong University. In the final round, held in Hawaii, teams raced to solve as many problems as possible within a five-hour time limit. Solutions in Pascal, C, C++, and Java were accepted. This year, the contest drew 3082 teams from 1300 universities worldwide. China and America dominated the final standings, with three teams from each country making it to the top 10; Russia, Canada, and Argentina were also represented. The U.S. colleges were MIT (second place), Stanford (fifth place), and Duke University (eighth place). Regional contests for the 2003 competition will begin in September. For more information, see http://www.acmicpc.org/.
The Department of Energy Science Grid (http://doesciencegrid.org/) was originally slated to be online in 2004, but is ahead of schedule and will become available later this year. The first computers to be connected will be three systems at the National Energy Research Scientific Computing Center (NERSC): the center's High Performance Storage System, its IBM supercomputer, and its Netfinity cluster. Systems at the Lawrence Berkeley, Argonne, Oak Ridge, and Pacific Northwest National Labs will join the grid in the future, allowing, for instance, fusion scientists around the world to share resources and advanced simulations. The Science Grid will use Condor tools (http://www.cs.wisc.edu/condor/) for resource management and the Globus Toolkit (http://www.globus.org/) for services. The SciDAC Collaboratory Middleware projects aim to solve the remaining problems; see a project list at http://www.osti.gov/scidac/computing/msdprojectlist.html.
Lockheed Martin's Joe Loughry and Auburn University's David Umphress have described a technique for reconstructing data from the activity of LED status indicators on electronic devices. This "optical TEMPEST" attack is capable of compromising "all data going through the device, including plaintext in the case of data encryption systems," they said in "Information Leakage from Optical Emanations" (http://applied-math.org/optical_tempest.pdf). "Experiments show that it is possible to intercept data under realistic conditions at a considerable distance. Many different sorts of devices, including modems and Internet Protocol routers, were found to be vulnerable."
Markus Kuhn, a graduate student at Cambridge, has documented a similar attack in his paper, "Optical Time-Domain Eavesdropping Risks of CRT Displays." Kuhn found that an attacker able to record the screen's reflection on a wall or a user's face can analyze the fluctuations in light intensity as individual pixels are lit to reconstruct the content of the screen. The method does not work with LCDs, in which all pixels on a screen are lit at the same time. Kuhn also compared his attack to TEMPEST, saying, "Very much like radio-frequency eavesdropping of video displays, the practical exploitation of compromising optical time-domain emanations eavesdropping will usually require specially designed equipment, expertise, and patience. However it seems at least as powerful as the former." Kuhn will present his paper at the IEEE Symposium on Security and Privacy.
The future of the venerable Xerox PARC, birthplace of the mouse and the desktop icon, remains uncertain. Xerox announced last year it would no longer fund PARC alone, and this year Xerox PARC officially became Palo Alto Research Center Inc. However, no other companies have expressed interest in backing the center.
Current research at PARC is focused on three areas: "Smart matter," which uses microscale technologies and optoelectronics; the "networks and documents" field, which covers the mobile office environment; and the "knowledge ecologies" group, which is concerned with text ana-lysis and transformation. For details on current projects, see http://www.parc.xerox.com/projects.html.
The most complicated problem yet posed to a DNA computer was solved at the University of Southern California. Dr. Leonard Adleman and his group solved an NP-complete, three-satisfiability problem with 20 variables, or 1,048,576 possible solutions. Previously, the most complex problem solved by molecular computers had nine variables — only 512 possible solutions.
The researchers assigned two strands of DNA, representing true and false, to each of the 20 variables. They then constructed 24 glass cells, each of which represented one possible combination of true and false for three specific variables, and placed the appropriate DNA strands in each cell. They then created over a million long strands of DNA, each representing a possible solution, and ran them through each cell. The long strands would bind to them and remain in the cell; others would pass through. At the end of the experiment, the strands that were caught by all of the 24 cells were known to represent the correct answer. For more information, see http://www.sciencemag.org/cgi/rapidpdf/1069528v1.
Gerard J. Holzmann, director of the Bell Labs Computing Principles Research Department, has received the Association for Computing Machinery's annual Software Systems Award for his work in automating ways for checking software for errors. According to Holzmann, typical error rates are about 10 errors per 10,000 lines of code. Holzmann's SPIN program, which he wrote about in his article "SPIN Model Checking" (DDJ, October 1997), enables speedy and efficient debugging and supports the formal verification of distributed systems.
The Rochester Institute of Technology has created a master's level concentration in game programming. Three courses currently make up the program: 2D Graphics Programming, Introduction to 3D, and 3D Graphics Programming. So far, 28 students from each program in the computing college — computer science, information technology, and software engineering — have signed up. Coursework is based on game-engine development with Microsoft DirectX. Driving the program, according to RIT, are facts such as: Sony Online Entertainment makes more money than Sony Pictures and, overall, computer-game programming is an estimated $20 billion industry and is predicted to grow to a $100 billion industry within a decade. Consequently, the demand for computer-game programmers is expected to increase.