Intelligent Machines

Synopsis: Information Technology

New publications, experiments, and breakthroughs – and what they mean.

Jan 1, 2005

Publishing for All
Democratizing contentpublication on the Internet

Context: Maintaining popular websites like Yahoo requires tremendous investment in bandwidth and powerful Web servers – investment that individual Internet users and small organizations can’t afford. If a small site’s content becomes extremely popular (as happens when a website like Slashdot links to it), its servers can become so overloaded that they can’t handle all the requests they receive. The power to publish popular content to large numbers of people on the Internet is thus restricted to large companies. A group of computer scientists from New York University recently put forward a system called Coral to remedy that situation.

Methods and Results: Coral allows one computer’s burden to be shouldered by many volunteers. In geek-speak, it is a decentralized and self-organizing peer-to-peer Web content distribution network. Users across the Internet volunteer their computers to collectively replicate and store the contents of popular websites. Internet surfers and Web page administrators can access or link to a website through Coral by adding “.nyud.net:8090” to its URL. A novel indexing technique allows Coral to quickly locate and retrieve the requested content. By distributing content so widely, Coral avoids high loads on both the original Web server and on the volunteer computers. A user is thus able to immediately access popular Web pages through the Coral network, even if the original Web server is reeling under heavy traffi c.

Why it matters: Coral offers the common Internet user large-scale publishing power. The new system distributes the server load across many nodes on the Internet and can easily handle any sudden spikes in demand for a particular website. That means users could host popular Web pages on their home computers over bandwidth-limited DSL or cable Internet connections without exceeding bandwidth or processing capabilities. Although Coral currently serves only static content and requires at least one Coral user to cache, or store, a web-site’s contents before its load spikes, the system offers the little guy a better chance of speaking to a big audience.

Source: Freedman, M., Freudenthal, E., and Mazieres, D. (2004) Democratizing content publication with Coral. Proceedings of 1st USENIX/ACM Symposium on Networked Systems Design and Implementation.

Keeping a Secret
A laser for quantum encryption

Context: Even the best communication security can’t prevent an unauthorized party from intercepting and attempting to decode a message. Quantum encryption harnesses a feature of quantum mechanics to solve this problem, making it impossible to observe (or tap into) a system without fundamentally disturbing it, and thus being discovered. Designs for quantum encryption systems have proven simple and elegant but so far impossible to build. Now a team led by Hong-Gyu Park at the Korea Advanced Institute of Science and Technology has moved one step closer to finding this Holy Grail, creating a microlaser capable of transmitting quantized light waves that may one day carry messages with greater security.

Methods and Results: The laser was fabricated out of the semiconducting material indium gallium arsenide phosphide, chosen because it can be fashioned to emit photons when a current passes through it. The critical lasing component is a “photonic crystal,” a perforated disc of the semiconductor that traps photons, ensuring that they are emitted at a single, constant wavelength. The crystal rests on a narrow post large enough to ensure good electrical activity but small enough not to disrupt the crystal’s structure. For the first time, the Korean researchers have demonstrated that an electrically activated microlaser can meet these competing needs.

Why it matters: For quantum cryptography to work, the creator of a message must be able to encode information in single photons and send them at set time intervals. Intercepting a photon destroys the information, revealing the presence of an eavesdropper. However, if a given bit of information requires more than one photon, the message can be intercepted without being detected. Consequently, the laser in a quantum encryption system must reliably convert an electrical pulse into a single photon at a prescribed wavelength. This work falls short of that goal, as it converts each electrical input into multiple photons. But it is an important step toward building new photon sources for optical communication.

Source: Park, H. G. et al. (2004) Electrically driven single-cell photonic crystal laser. Science 305:1444-7.