Hulya Seferoglu awarded CAREER grant to research edge computing

Assistant Professor Hulya Seferoglu was awarded a National Science Foundation Faculty Early Career Development Program (CAREER) grant this spring for her research on coded computation in distributed edge computing systems.

As an ever-increasing number of Internet of Things (IoT) devices such as wireless sensors, smartphones, robots, Fitbits and self-driving cars come online, the need for distributed computing, or processing data near the end device, becomes more critical.

Unlike cloud computing, where information is processed in centralized data centers that can be far removed from the end user, processing happens nearby in edge computing. Multiple devices close to end users pool power to process data simultaneously and in a distributed way. These devices can include smartphones, tablets, routers, and base stations.

By way of example, Seferoglu said a single self-driving car generates roughly 10 gigabytes of data per mile driven (to put this in perspective, you can watch 10 hours of Netflix using the same amount of data). Transmitting that data from the car to the centralized cloud requires a large amount of bandwidth and will not result in timely processing.

“In our current infrastructure we cannot send this amount of data. You need to use either cellular or WiFi and the current systems don’t support this,” Seferoglu said.

Seferoglu’s CAREER grant will allow her to look at practical applications of coded computation, a theoretical framework, to the recent area of edge computing. Coded computation is a field of coding theory that uses erasure codes, or error-correcting codes, through data redundancy, in distributed computing systems. She aims to characterize the cost-benefit tradeoffs of coded computation in edge systems, and develop networking algorithms and protocols to make the framework adaptive to the nature of edge computing systems.

“How can we develop practical distributed computing mechanisms to harvest heterogenous resources at the edge? And how can we develop mechanisms to exploit these resources?” Seferoglu asked. “Privacy and security are more challenging in edge computing as well. We are looking at security, efficiency and privacy.”

She has spent her entire career in network coding, both the practical aspects of network coding and theoretical work on information theory. While the idea is similar, Seferoglu will be examining how to use coding mechanisms in a practical edge computing system.

“Things are changing and edge computing is becoming more important, it’s one of the most important topics in the networking community right now. I thought, how can we use this in practice?” Seferoglu said.

The five year, more than $500,000 award, titled “CAREER: Practical Coded Computation Mechanisms for Distributed Computing,” will support one PhD student for five years. Seferoglu, who has been at UIC since 2013, has several active grants, including a $900,000 study with the Army Research Lab, and a National Institute of Standards and Technology grant for $270,000.

To learn more about Seferoglu’s work, visit the Networking Research Lab.