This year, UCLA Computer Science professors Jason Cong and George Varghese were elected as two of the 84 new members and 22 foreign members of the National Academy of Engineering’s latest class of scholars. Founded in 1964, the National Academy of Engineering is a private, nonprofit institution that provides engineering leadership in service to the nation. Its members consists of senior professionals in business, academia, and government, who are among the world’s most accomplished engineers. Election to the NAE is among the highest professional distinctions an engineer can earn, and this year, with the election of Cong and Varghese, the UCLA Computer Science department was the only department at any university to have more than one elected member.
“This is a tremendous honor,” Cong said. “It’s a very highly distinguished group, and it’s great to be a part of that. I think this is also recognition for the outstanding work of my graduate students.”
Cong received his B.S. in computer science from Peking University in 1985, and his M.S. and Ph.D. in computer science from the University of Illinois at Urbana-Champaign in 1987 and 1990 respectively. He joined the UCLA computer science department in 1990, where he focused primarily on researching field-programmable gate arrays (FPGAs) and developing the now well-known FlowMap algorithm. In his research, Cong proposed that the inefficiency of general purpose processors could be circumvented by instead using a totally customizable logic via FPGAs. FPGAs – basically, integrated circuits that can be reprogrammed for various dedicated functions in less than a second – allow for the sort of customization that would increase efficiency, but figuring out how to program something of this kind is extremely difficult. Cong’s research made FPGA programming much easier and more efficient. One example is his early work on the FlowMap algorithm, which presented a theoretical breakthrough by developing a polynomial-time depth-optimal algorithm for mapping a general Boolean network to a network of programmable logic blocks. Another example is the high-level synthesis work by Cong and his team, which can compile C/C++ programs into FPGAs with automated scheduling and resource allocation, as opposed to using special hardware description languages for circuit designs, resulting in a much improved design productivity. It was these “pioneering contributions to application-specific programmable logic via innovations in field-programmable gate array synthesis” that were recognized by the National Academy of Engineering.
Cong’s passion for algorithms started at a young age. Early on, he had a natural curiosity for mathematics, and during the 1980s, he began to take interest in the developing technology of the time, fascinated by how machines were capable of doing calculations even faster than the human beings that created them. His interest in researching algorithms began to take root when he was working on a programming assignment during his undergraduate career in Peking.
“There was a straightforward way to do [this assignment], and a clever way. I actually used the clever way, while most of my classmates used the straightforward way. At the time, we were using a computer that used punch cards as input – so, when the machine printed out the result, you could actually hear the difference in efficiency. The faster algorithm printed the result in seconds, while the straightforward algorithm was left printing [for a while],” Cong said. “I could really see the power of different methods of computation, and that got me interested in studying even more efficient algorithms”.
Cong says in the long run he hopes to achieve energy efficient computing beyond the commonly used Von Neumann architecture. “Energy has become a serious bottleneck in computation,” he said. “We can think about energy efficient computing when we look at the human brain. Our brains are capable of accomplishing a lot of things, but we only use maybe around 20 watts,” Cong said. It’s this kind of architecture that Cong says we should model our computing after, and the use of FPGAs is a “significant way to approach this sort of efficiency.”
George Varghese, also a Chancellor’s Professor of Computer Science, was recognized for his work on “network algorithmics that make the Internet faster, more secure, and more reliable.”
“I’m just grateful,” Varghese said, when asked about his thoughts on receiving the election. “It’s based on work done with lots of students, so I’m grateful for what I would call God’s grace in connecting me to these problems and people.”
In 1981, Varghese received his B.S. in electrical engineering from the prestigious Indian Institute of Technology, Bombay, and in 1983, he received his M.S. in computer studies from North Carolina State University. In 1993, Varghese received his Ph.D. in computer science from the Massachusetts Institute of Technology. After receiving his Ph.D., Varghese felt he had a calling to teach, and went on to lecture at a number of universities like Washington University, Stanford, and UCSD, until finally joining the UCLA Computer Science Department in 2016.
Varghese’s early research focused on speeding up network devices via network algorithmics. He uses the phrase “algorithmics” to express the idea that his work involves not just algorithms, but also viewing the problem holistically – taking into account the whole picture, and understanding the impact of a number of different factors, like the capabilities of the physical hardware.
“Whenever you send a message on the internet, it goes through something called routers, which send things through one router to another until it reaches its destination,” Varghese explained. “There was a period of time where the physical links were getting faster and faster, but the routers themselves were slow. I spent most of my early research on trying to make the forwarding processes of the routers as fast as the physical links.”
Varghese compiled his research into the now well-known publication, “Network Algorithmics: An Interdisciplinary Approach to Designing Fast Networked Devices”, where he details his complete methodology for maximizing the speed of network devices. Several of the algorithms he helped develop are now widely accepted and used by companies like Cisco, Juniper, and Microsoft.
Varghese’s current research with UCLA Professor Todd Millstein focuses on network verification – treating networks as mathematical objects and building tools to either prove properties about large clouds, or discover bugs as counterexamples. “If you make a single mistake in a network, it can bring down the network for hours,” Varghese says. “Humans make a lot of mistakes when they configure the network, so we’re trying to use formal methods to detect these bugs early on.” He hopes his work will help in developing more secure and more reliable networked software. “When you’re building things like bridges and airplanes, you can’t afford to make a mistake, but when a network crashes, it’s only a mild annoyance – but, eventually we want to make the network as reliable as these structures.”
With the election of Cong, Varghese, and Electrical Engineering Professor Behzad Razavi, UCLA Engineering has 35 affiliated faculty members who are National Academy of Engineering members. Cong and Varghese will be officially inducted on Oct. 8 during a ceremony at the NAE’s annual meeting in Washington, D.C.