Photo by Evan Krape | Illustration by Joy Smoker
July 06, 2022
Lena Mashayekhy, assistant professor of computer science at the University of Delaware’s College of Engineering, is taking a proactive approach to how artificial intelligence and analytics can be used to safely and swiftly process vast amounts of data across devices as we begin to interact with more and more types of technologies in our daily lives.
The so-called “Internet of Things,” devices that help our world become more connected — smart watches and other wearable objects, drones and autonomous vehicles, for example — is expanding and there’s a need to anticipate future mobility challenges to real-time decision-making.
“In today’s world, we are moving toward many data-driven, decision-making applications,” Mashayekhy said. “In recent years, devices that are at the edge of the internet, they generate a lot of data, such as images and videos, and need to make real-time decisions. The significant increase in both Internet of Things network size and data volume opens up attractive opportunities for data analysis and learning to extract valuable information to support decision making and derive scientific discovery and innovations.”
Her forward-looking thinking and research has earned Mashayekhy recognition from the National Science Foundation (NSF) in 2022 with a Faculty Early Career Development Program (CAREER) award, one of the most prestigious awards a junior faculty member can receive.
The award includes $671,835 in funding over a five-year period starting May 2022 to support this project.
“Just like desktops and phones have transformed our lives over the past few decades, countless small computing devices in almost any conceivable object will likely assist our future existence,” said Rudi Eigenmann, interim chair of the Department of Computer and Information Sciences. “Professor Mashayekhy’s research promises to become a critical enabler of this development, empowering us on the move — when walking, driving a car or even flying an airplane.”
Think about how an autonomous car would need to sense a person crossing the street or how a search drone would need to communicate that it had found a missing person — these decisions need to be made immediately, within seconds. That requires serious processing power to capture an immense amount of data, analyze it and turn all of that information into an actual decisive response.
While some control operations, such as maneuvering, route planning and avoiding collisions of autonomous vehicles, for example, have to be performed on the device itself, other time-critical analyses require powerful computing processing. For instance, surveillance drones with applications in search and rescue, defense and smart cities require searching large areas and analyzing high-definition images and videos to quickly identify their targets (such as an injured person or a missing child). Such devices must also be able to assess their target’s status for important information such as signs of cardiac events, risk of heat-related injuries and air contamination threats.
“This kind of multiaccess edge computing enables Internet of Things devices to offload their delay-sensitive resource-intensive tasks. As we or these devices move, our connection to those computational resources will change, but the devices should not experience any interruptions,” she said. “I’m planning to explore how to deal with mobility challenges proactively and provide stable computing services.”
The issue here is much more impactful than simply solving the loading lag while trying to upload photos or videos to social media.
“These days and in coming years, everything will be connected,” said Chien-Chung Shen, a professor in the Department of Computer and Information Sciences who served as the chair of the recruiting committee when Mashayekhy was hired. “There’s a lot of processing that needs to be done. You want to minimize the delay, even though all of these decisions are subject to real time constraints. All of this needs to have very advanced resource planning.”
The Internet of Things is driving us toward a much more digitally reliant future. By using multiaccess edge computing, which allows the necessary data processing closer to the user’s devices, Mashayekhy and her team of researchers hope to improve the quality of connections and computing services within the fast-paced mobile world we live in today.
“Many of these problems have been formulated into optimization problems,” Shen said, noting that Mashayekhy is a well-respected expert in this field of study. “The impact will provide insights and critical solutions that help the carriers — Verizon, for example — to deploy their edge devices.”
The proposed project focuses on mobility challenges from three different perspectives. Mashayekhy and her lab will first evaluate how user movement creates challenges, and how those problems could be addressed.
Next, they will tackle challenges caused by infrastructure mobility, where an edge device would be mobile instead of stationary. For example, a larger drone could provide the localized computing power needed by search drones while they do their important job.
Lastly, the project will explore incentive programs for users that could help balance the load of an entire digital system. For example, when electric companies advise people to “beat the peak” and reduce high-energy consumption during the hottest parts of the day in the summer, there are also benefits to having digital users spread out demand. It could apply to offering incentives to autonomous vehicle users to take a different route than the busiest most popular one, for example.
“In this project, we will investigate all aspects of mobility in multiaccess edge computing,” she said. “This project also looks at how we can drive ideas from deep learning, game theory and physics, like center of gravity and Newton’s law of universal gravitation, rules that impact maintaining balance and adjusting drone-based edge devices in response to dynamic and uncertain Internet of Things device mobility.”
NSF CAREER Awards blend research and education, and Mashayekhy’s project will enlist two doctoral students to assist with research. She said there will also be opportunities for undergraduate students to participate in research during the five-year project.
The project also will support an integrated educational program and outreach activities for K-12 students, their families and the broader community to expand exposure to this research while also empowering more innovative applications that are not yet possible today.
“This recognition by NSF is an acknowledgement of not only her robust and exciting research program, but also her exceptional record as a strong and dedicated teacher and mentor who can pass along the excitement of this research to others,” said Kathleen McCoy, a professor and former chair of the Department of Computer and Information Sciences. “Her research fills a critical area in computer systems — a very important research thrust in the department — and it is also important to many interdisciplinary research efforts within the College of Engineering and across the University of Delaware, such as efforts in computation needed for autonomous vehicles.”