News Release

Taming internet complexity for a more responsive user experience

A research team at UMass Amherst has received a four-year, $2.4 million NSF grant to study the increasingly complex ways in which content is delivered on the Internet and to invent new architectural and algorithmic mechanisms to coordinate these better.

Grant and Award Announcement

University of Massachusetts Amherst

Ramesh Sitaraman, University of Massachusetts at Amherst

image: Sitaraman and colleagues want to influence the future evolution of all organizations within what they call the Internet content ecosystem view more 

Credit: UMass Amherst

AMHERST, Mass. – A research team at the University of Massachusetts Amherst has received a four-year, $2.4 million National Science Foundation grant to study the increasingly complex ways in which content is delivered to users on the Internet and to invent new architectural and algorithmic mechanisms to coordinate these better.

Computer scientists Donald Towsley and Ramesh Sitaraman of the School of Computer Science, with electrical and computer engineer Christopher Hollot at UMass Amherst and University of Minnesota computer scientist Zhi-li Zhang, aim to influence the future evolution of all organizations in what they call the Internet content ecosystem.

Providing content to users around the world in the form of web pages and videos is the dominant function of the Internet, which involves multiple organizations, the researchers say. Content providers such as Netflix, Google, NBC or Facebook have the content users want. Content distribution networks (CDNs) such as Akamai distribute that content to users. And Internet service providers (ISPs) such as AT&T own and operate the underlying network infrastructure.

"The Internet used to be a simple place, but no longer," Sitaraman says. "Even the seemingly simple act of a user watching an online video triggers a complex series of interactions between the content provider providing the video, the ISP providing Internet connectivity to the user, and one or more CDNs tasked with delivering the content. These interactions are often loosely coordinated and can sometimes go unexpectedly awry, leading to an inferior online experience."

The key to the future success of the Internet rests squarely on understanding these complex interactions and in inventing new architectural and algorithmic mechanisms to coordinate them better, the researchers say. But this is no easy task because the Internet consists of thousands of organizations that often act independently with minimal coordination.

"It's a little like having many cooks trying to cook the same dish," says Sitaraman. "Consider how the quality level at which a video should be streamed to a user is decided. The video player picks a quality level based on its estimation of the bandwidth that the network can provide, while the network is using its own protocols to independently divide its bandwidth resources across multiple users. The lack of coordination in the decision-making can result in anomalies such as users seeing low quality videos even when the network bandwidth is plentiful."

Towsley adds, "Solving the puzzle requires a truly interdisciplinary approach. We will apply a rich set of techniques from control theory, network algorithmics and mathematical modeling to develop new and better ways of managing complex interactions between these organizations."

Hollot says, "The scientific knowledge and tools developed from our research will help design next-generation CDNs and ISPs that provide a better quality experience to users, which will in turn result in better monetization outcomes for content providers."

The researchers also plan to include undergraduate and graduate students, in particular women and underrepresented minorities, in their research activities and conduct outreach activities to engage with the Internet industry.

###


Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.