Developing Tomorrow's Internet
UNM team innovates a more secure and mobile Internet
In early February, four undersea Internet cables in the Middle East failed, one severed by a six-ton ship anchor. Outages across the region and to American military forces in the area were reported and Internet traffic was snarled for days. It isn't uncommon for undersea Internet cables to break or go offline, but the fact that four major cables serving one region failed at the same time emphasized the vulnerability of this critical communication network.
The FIND Challenge
That scenario is a perfect example of why research teams from around the United States are working on the Future Internet Network Design, an initiative sponsored by the National Science Foundation. FIND challenges researchers to move beyond the Internet's current constraints and redesign it to address questions about security, information dissemination, economics, and preserving a free and open network. "This project asks us to totally forget what we have now, and instead, see what we would want to do if we built the Internet from scratch," explains Chaouki Abdallah, chair of the Electrical and Computer Engineering Department. "We also need to imagine how people will use the Internet fifteen years from now and build it to accommodate those new uses and expectations." Abdallah is working on the FIND project with Wei Wennie Shu, associate professor of electrical and computer engineering; Greg Heileman, professor of electrical and computer engineering; and a number of UNM master's and doctoral students.
Why does the Internet need to be reinvented now? When Robert Kahn and Vinton Cerf developed the data transmission standards that are the backbone of the current Internet, their goal was to keep a few computers connected in case of a nuclear attack.
Today, it's a very different story. Hundreds of millions of computers, cell phones, and PDAs are connected through the Internet, and the network is used for everything from selling products to making friends and launching malicious attacks. "The Internet, as it exists today, is very clumsy and slow because it was designed for hundreds of things, not billions of things," says Abdallah. "The problems of security, wireless access, hackers, and spam have grown exponentially. Our research addresses these issues and works to support what users will do with the Internet of the future."
In 2006, UNM received a two-year $500,000 grant from NSF. UNM is splitting the grant with its research partner, the Corporation for National Research Initiatives (CNRI), a non-profit organization that fosters research on the national information structure. Internet pioneer Robert Kahn is president and CEO of the organization. "It has been very rewarding for our UNM team to work with Dr. Khan. His insights into the original Internet design have been very valuable," says Abdallah.
The team is thinking outside the box - indeed far beyond any known framework - to address all challenges related to developing tomorrow's Internet. While most FIND research teams are looking at a single aspect of the Internet redesign such as security or mobility, the CNRI/UNM team is one of only two working on all aspects of the Internet's new architecture. "The work we're doing cuts across every level of the project from the physical infrastructure and how networks form, all the way to changing social attitudes and social networks," explains Abdallah.
Network Architecture
Using a variety of mathematical tools including control theory, game theory, network science, simulation, and programming, the CNRI/UNM team is re-engineering the Internet as a logical network, not a physical one. They call their approach transient network architecture and it resolves two of the biggest Internet design challenges. "The transient network architecture emphasizes mobility and at the same time addresses security. From the very beginning of the design process, we've considered both factors," says Shu. "Within such a logical framework, it is not only about communicating information, but also about managing it," explains Heileman. "While the original Internet architecture was tailored for machine-to-machine applications, today's Internet usage model is centered around information access and retrieval."
Most of the functions currently dependent on hardware, like computers and PDAs, and physical infrastructure, like routers and undersea cables, would be transferred into software capable of migrating when resources are low, hardware fails, or cables are cut. Using control theory to evaluate its best resource options, the software would migrate to another working node - another computer or cell phone, for example - and the system would reallocate resources accordingly to keep content available and functions running smoothly. "We've designed this so if there is a problem with the network, the software moves because what you're concerned with is not actually the hardware but rather the information and the ability to keep things moving forward," explains Abdallah.
To appreciate how convenient transient network architecture would be, consider what happens when you lose your Blackberry. Right now, when you leave your PDA on a plane or drop it in the ocean, you've lost the hardware and all the valuable - and sometimes sensitive - information on it. With transient network architecture, you could simply call the Blackberry from another device, enter a few commands, and transfer your address book, emails, and other personal information to a new device. You'd have what's most important to you - your information - and the Blackberry device itself would essentially become a useless shell.
Unique Identifiers and Associations
Part of what makes this transient Internet concept possible is the team's use of persistent identifiers (PI), unique numbers attached to everything associated with the network. So regardless of where it is in the network, every person, computer, document, or software file would be assigned a unique PI, much like a social security number.
Eventually, everything from people and equipment to appliances and clothes with sensors in them would have a PI. "In the future, networks are going to consist of huge numbers of sensors and everything will have a PI and need to be networked," says Heileman, whose focus is digital rights management and content management. "Persistently and securely identifying network objects, whether users or content, enables a powerful notion of secure identity in the network. Our research builds upon this powerful notion to design an open, flexible information and digital rights management architecture that can better address the needs of a future Internet."
PIs would replace the current system of IP addresses, which are essentially ever-changing geographical locators associated with only a piece of hardware. Henry Jerez, a senior research scientist at CNRI and a UNM SOE graduate, studied how PIs are currently used by libraries to manage publications and digital objects. Then he adapted the idea for the UNM/CNRI FIND project so that PIs can be applied to computers, cell phones, people, and more. Joud Khoury, a PhD student at UNM, is one of the main contributors to the project. His dissertation work, in addition to investigating scalable naming architectures, is providing a taxonomy of most known networking architectures, and will, for the first time, allow for a fair comparison between proposed new architectures.
To manage the process of connecting so many PIs and the ad hoc way networks tend to form, the UNM/CNRI team created areas of influence (AOIs), distinct levels of connectivity between PIs within a specified region and among different AOIs. There are three AOI circles: local, intermediate, and global. When one PI tries to connect with another, the AOI system first searches for the two PIs at the local level. If they aren't present at that level, the system expands the search to the intermediate level and then, if needed, to the global level.
Now in their second year of the FIND initiative, the UNM/CNRI team has already tested their new transient network architecture on the UNM campus using a robot, PDAs, and cell phones. The robot can move from one part of the campus to another, disconnect itself, and then connect at a remote location seamlessly. This test effectively demonstrates that devices and agents may connect to the Internet from anywhere in the world, be able to receive commands, and coordinate their actions with nearby or remote devices. Now the team is hoping to test the system on the City of Albuquerque's Rapid Ride buses.
Twice a year, the team meets with other FIND teams to share results and ideas. Ultimately, the best research results from FIND will be tested through the Global Environment for Network Innovations (GENI). GENI is a NSF-sponsored research facility designed for realistic experiments on the radical network designs of today that could become the design standards of tomorrow.