Electrical Engineering Assistant Professor Alphan Sahin has received a National Science Foundation (NSF) CAREER Award to study the use of wireless signals in computation applications. His research is expected to establish over-the-air computation as a means of faster computation for decision-making, learning and coordination.
As principal investigator of the five-year, nearly $500,000 project, Sahin will lead his team in testing a wireless medium as a computational substrate to perform calculations in various applications by harnessing the additive nature of electromagnetic waveforms, which result in interference.
“Wireless signals have long been a cornerstone for many communication-based applications, from mobile cellular systems to navigation, where we typically do not want interference,” Sahin says. “This project explores using interference in a useful manner to use wireless signals from multiple sources simultaneously for computation.”
The research team will build a software-defined radio platform to test whether computation through superposed signals through interference is handled accurately. Sahin expects the work will improve the understanding of low-latency computation over wireless networks resulting in faster autonomous systems, technology that would benefit multidisciplinary industries such as networked robotics and vehicular networks.
Societal impacts include noticeable latency improvements in computation-heavy applications that affect daily life, ranging from traffic flow to environment monitoring and cost reduction through better spectrum utilization.
Electrical Engineering graduate student Anthony (AJ) Perre serves on Sahin’s research team. He mentions the majority vote as one application for over-the-air computation.
“Imagine you have a large group of people, and you want to tally votes. Typically, you would gather individual votes and determine the majority after collection,” Perre says. “But imagine if there was a way everyone could shout their answer at once, and by nature of audio signals, I could understand the majority.”
Sahin describes storage containers at an Amazon warehouse as another example where over-air-computation could be beneficial. If workers needed to detect the average temperature in all of the containers, rather than checking each individual sensor, signals could be transmitted simultaneously, with the average computed using this technique. The same technique can also be used to aggregate distributed neural network parameters as quickly as possible over wireless networks.
“Dr. Sahin’s project is extremely relevant for current and future wireless systems and networks,” says Mohammod Ali, chair of the Department of Electrical Engineering. “I firmly believe his research will break new grounds to increase our understanding of the fundamentals of low-latency communication and find solutions for computation-heavy wireless applications like networked robots, vehicle networks and other faster autonomous systems. His project has a well-developed plan that integrates education with research, benefiting our students at both the undergraduate and graduate levels.”
Each NSF CAREER Award project must integrate research, education and outreach. Sahin’s project includes a comprehensive action plan designed to train students, engineers and researchers on relevant problems in communications, computer science and mathematics.
The team will facilitate outreach to high school students through collaboration at educational conferences and publishing research papers. A YouTube channel will be developed to reach and educate a more widespread audience on wireless communication applications.
At the university level, Sahin plans to expand undergraduate education within the Department of Electrical Engineering through coursework and research opportunities in the intersection of wireless communications, signal processing and mathematics.
“We want to enhance wireless education within the department, train students who understand not only wireless communications, but also its impact on artificial intelligence systems, and teach them how to use wireless networks and radio signals efficiently to facilitate the underlying communications for artificial intelligence systems,” Sahin says.
Through this project, he will train students in using radios to experiment with wireless signals and sponsor an undergraduate Capstone project focused on neural networks.
“Giving students more exposure in areas related to wireless systems is a good thing,” adds Perre. “We take for granted how much the world relies on wireless signals and their role in the Internet, communications and things that are fundamental to our daily lives. As students, we need more exposure to these systems that we use every day.”