August 31, 2010 | 9
The middle-aged Internet (ARPANET first went live more than 40 years ago) could easily slide into complacency, but the National Science Foundation (NSF) might be staving this off with four multimillion-dollar grants that the agency has recently awarded. The Future Internet Architecture (FIA) research projects are expected to re-think the network from the ground up, taking into account emerging security concerns, the demand for greater bandwidth and the growth of mobile devices.
The Internet’s original design accounts primarily for information to be passed from one host server to another along a wired network. Attempts to secure these hosts and networks have come as an afterthought (ARPANET was originally a closed network) and have struggled to keep pace with society’s expanding economic and social reliance on the Internet. Likewise, the host-based architecture (where computers seek access to information from a specific server or group of servers) is starting to look creaky as the number of computers and mobile devices seeking access multiplies exponentially each year.
An NSF-sponsored FIA project headed by Lixia Zhang, a computer science professor at the University of California, Los Angeles, seeks to create a more efficient Internet through the distribution of data. Zhang and her team are developing what they call a Named Data Networking (NDN) architecture. In a November 2009 presentation at Keio University (pdf), Zhang used YouTube as an example of how this would work. Under the Internet’s current architecture, if a popular video is downloaded 1 million times, that means 1 million download requests are sent to the same YouTube server (pdf). Under NDN, a YouTube request would find a nearby copy of the video rather than requesting it directly from YouTube’s (or Google’s) data center. The challenge here, since you’re not dealing with YouTube directly, is making sure you’re downloading the video, as opposed to a piece of spyware dressed up to look like that video.
Another FIA project, led by Peter Steenkiste, professor of computer science and electrical and computer engineering at Carnegie Mellon in Pittsburgh, poses the creation of a more secure eXpressive Internet Architecture. XIA is similar to the NDN architecture in that both propose to secure data itself rather than the network pathway that the data travel. XIA likewise promises to help users find content wherever it is most accessible, even if this isn’t a host server, speeding information retrieval while cutting redundant network traffic.
Improved security goes hand in hand with the use of a more distributed infrastructure to store data for retrieval via the Net. Under XIA, a document will have a specific identifying tag to verify that the document you’re getting is the document you were looking for, Steenkiste says. Think of this in terms of the kind of authentication services that VeriSign, Inc. offers, except the authentication would be built into the Internet rather than added on later. The researchers expect to have a prototype XIA-based network ready for testing within a year. NSF is giving the XIA project $7.1 million over the next three years.
NSF is also sponsoring an FIA project that redesigns the Internet to be more tolerant of roving mobile communications, which can originate from a variety of locations and are often interrupted due to dropped signals. The $7.5 million MobilityFirst project is being led by Dipankar Raychaudhuri, a Rutgers University professor of electrical and computer engineering and director of the school’s Wireless Information Network Laboratory in North Brunswick, N.J. Beyond mobile phones, MobilityFirst seeks to support a variety of different types of communications, whether they are signals between health monitors and hospital equipment or connected cars sharing information to help avoid collisions.
The fourth FIA project focuses on the growing use of data centers to store software and information that used to be stored locally on PCs (the "cloud computing" model). Led by Jonathan Smith, a University of Pennsylvania professor of engineering and applied science and of computer and information science, the Nebula architecture (pdf) (to be supported by NSF to the tune of $1.76 million) will be designed to facilitate the use of data centers connected by a high-speed, highly reliable and secure backbone network as the primary repositories of data and sources of computation.
The original Internet had one great advantage over any of its successors. It was launched without fanfare to meet the specific information-sharing needs of a very specific audience—scientists working for the Defense Department. Any decisions made about the Internet must now satisfy a much greater number of parties, in particular the software, networking and hardware companies whose technology made the network what it is today. It may be difficult to reach a consensus about the way forward, but these NSF projects should be helpful in providing a number of options.