Feature Stories


UMass Amherst experts are driving the future of Internet infrastructure
  • Tablet devices with internet icons hovering above hand.

“If mobility is the norm, we really want to think about building a network architecture that takes mobility as a first-class concern and organizes the whole architecture around it.”

—Jim Kurose

Our current Internet architecture was built for a wired world and has not kept pace with the rapid explosion of technical, business, and social applications. This puts users at risk for security breaches and exposes discrepancies between old, outmoded capabilities and an increasingly mobile user-base. Federal initiatives are beginning to address these issues with high-level competitive research efforts, in which two out of the five major projects find UMass Amherst researchers at the helm.

The workup to these efforts began in 2002 at a National Science Foundation–sponsored workshop designed to explore issues surrounding outdated Internet architecture, including the need to explore new technologies. Chaired by UMass Amherst Distinguished Professor James Kurose (computer science) and attended by researchers from around the country, the workshop yielded a report that served as a springboard for federal research programs aimed at addressing next-generation Internet architecture needs. The newest and largest of these is the Future Internet Architecture (FIA) Program, which is funding five multi-million-dollar projects nationwide. In two of these, UMass Amherst computing and engineering experts Kurose, Arun Venkataramani, Tilman Wolf, and Michael Zink are key players. Along with world-renowned UMass operations management expert Anna Nagurney, they are part of a core faction of researchers helping to create the next-generation Internet.

“The fact that our campus was selected to take a leadership role on two of the five FIA projects speaks volumes about our faculty expertise,” says Mike Malone, vice chancellor for research and engagement. “We take an interdisciplinary approach to science, which is critical to an initiative of this size.”

Kurose, Venkataramani, and Zink are part of FIA’s MobilityFirst project, led by a team at Rutgers University and aimed at designing an alternative Internet architecture with mobility and security as the driving goals. When the Internet was standardized there were no mobile devices, yet now such devices are in the majority. Kurose explains that it makes little sense to have such an important network designed to connect with outdated technology.

 “If mobility is the norm,” he says, “we really want to think about building a network architecture that takes mobility as a first-class concern and organizes the whole architecture around it.”

 With Venkataramani as lead architect, the UMass team is working with colleagues at Rutgers, the University of Wisconsin-Madison, the University of Michigan Ann Arbor, Duke University, MIT, and the University of Nebraska-Lincoln to build the overall architecture, put protocols in place to substantiate the architecture, and develop working prototype code.

According to Venkataramani, issues of mobility and security can be addressed simultaneously by separating names, or endpoint identifiers, from their addresses, or network locations. By conflating the two, he says, the current system poorly supports mobility and leaves IP addresses vulnerable to hijacking. A key innovation in MobilityFirst is the Global Name Service, or GNS—software designed specifically to support mobility in a seamless, secure manner.  Logically centralized but physically geo-distributed, GNS is a cloud-based infrastructure that maintains information about people and their associated locations and enables “context-based communication,” meaning that it generalizes name- or address-based communication. In developing the technology, the team realized its potentially revolutionary implications for notification systems around the globe. Because the software stores basic identification information, can pinpoint one’s various mobile devices (phone, iPad, or even a car GPS), and can send contextualized messages, it is easy to imagine how it could alert people traveling in a certain direction of an imminent traffic obstruction or even a developing tornado. It can also customize different messages to different people—for example, one warning to senior citizens and a different one to first responders.

Wolf and Nagurney are leading another FIA project, ChoiceNet, an Internet “economy plane” designed to provide greater economic incentives for service providers and more options for users. Imagine sitting down to a movie marathon on a rainy Sunday with your family and purchasing, for that day only, high-speed, high-definition streaming options. Or imagine working from home while your office is being renovated, and purchasing high-speed options for just two weeks rather than for a whole month or more. Wolf and Nagurney say that such options would be more convenient for users and would force competition between service providers, leading to better technologies overall and more competitive prices.

“Our project is about giving people choice,” says Wolf. “We’re trying to think about how we can restructure Internet architecture so that we can actually incentivize providers to deploy new and innovative protocols and services in the network.”

Wolf began thinking along these lines during a previous NSF project, in which he attempted to improve the functionality of routers within the Internet network. He and his team felt discouraged knowing that their work would not readily be deployed under the current model because providers had no incentives to spend the money to improve the technology. ChoiceNet, he explains, opens new doors for network-based services and builds economic relationships for services across various time scales.

Wolf, the project’s PI, is leading a team across the UMass Amherst, UKentucky, NC State, and UNC campuses. An engineer, he partnered with Nagurney to ensure that the technology he is developing includes the right economic incentives. Nagurney and her graduate students are investigating the behavior of various stakeholders in order to better understand the competitive environment and its pricing, service-quality, and provider-profit implications.

Zink, also a co-director on the campus’s CASA (Collaborative Adaptive Sensing of the Atmosphere) Engineering Research Center, is working with other colleagues on a subsequent grant to layer their advanced radar technology over the GNS to enhance its capabilities. As a planned field trial, they will work in coordination with the National Weather Service and the CASA radar test-bed in Texas to demonstrate the effectiveness of context-based hazardous-weather warning apps.

One important aspect of the network being worked on in all FIA projects is caching—finding ways to more efficiently utilize storage. In the current model, the vast majority of storage and computation happens at the edge of the network, while the hardware inside the network is relatively slow and storage-poor. To that point, NSF’s Global Environment for Network Innovation (GENI) program was launched along with FIA as a way to test new architectures with a more cutting-edge network that implements advanced hardware. Because the Internet serves billions daily, it has been difficult to experiment with alternative Internet architectures while in “production mode,” which Kurose likens to “changing the engine of an airplane while it’s flying.”

The GENI program addresses that issue by funding projects to help build a smarter, faster experimental platform on which to test the FIA projects and other Internet-advancing projects around the globe. The platform is a working infrastructure strictly for research traffic, not commercial. In a project closely aligned to GENI through an NSF CC-NIE (Campus Cyberinfrastructure – Network Infrastructure and Engineering) grant, Zink says that he and his colleagues will soon install new switches to connect UMass Amherst with the Massachusetts Green High Performance Computing Center in nearby Holyoke, a data center dedicated to research computing.

A platform called NSF-Net predates GENI, having been used since the 1980s, before the Internet was commercialized in 1993. Now that the Internet is used by so many on a day-to-day basis, Zink explains, the industry is more likely to insert the GENI platform’s more advanced aspects into the current model than to switch to an entirely new system.

“I think they’ll adopt whatever’s interesting for them,” Zink says.

Amanda Drane '12