Georgia Tech Partners with USC for $6.8 Million DARPA Project
Researchers teaming to develop software toolkit for revolutionary next-gen HIVE processors
(Georgia Tech, Atlanta, GA, 10 July 2017) The Georgia Institute of Technology and the University of Southern California Viterbi School of Engineering have been selected to receive Department of Defense Research Projects Agency (DARPA) funding under the Hierarchal Identify Verify Exploit (HIVE) program. Georgia Tech and USC are to receive total funding of $6.8 million over 4.5 years to develop a powerful new data analysis and computing platform.
Many security and consumer applications – including identifying and zeroing in on erratic driving behavior of vehicles in real-time, recognizing terrorist cells through patterns of communication, or protecting critical infrastructure facilities such as power, communication and water grids, or even predicting the spread of a cyber attack – can be modeled using graph data-analysis formalisms envisioned in the HIVE program.
Georgia Tech and USC will be responsible for developing a software toolkit to work on HIVE processors being developed by hardware vendors. The goal is to process data at a rate 1000 times faster than existing hardware and software techniques.
Georgia Tech’s School of Computational Science and Engineering Chair David Bader will lead the academic development project, and the two universities will co-develop software to quickly process the incredible amount of data from cellphones, social media and other sources, and demonstrate the relationships among data points in real-time.
“With the research infrastructure and capabilities at Georgia Tech – including the Center for Research into Novel Computing Hierarchies – and at USC, we are well-poised to deliver a robust yet stable software platform that will allow future programmers to fully leverage the revolutionary capabilities and performance that HIVE processors are expected to provide,” Bader said.
Viktor Prasanna, professor of electrical engineering and a professor of computer science at USC Viterbi School of Engineering whose research interests include high-performance computing (HPC), hardware-software co-design and data science, will head the program from USC’s end. Rajgopal Kannan, a research associate in USC Viterbi’s Department of Electrical Engineering will also collaborate on this project. With years of network, architecture, and HPC experience, Research Scientist Oded Green will also play a key role at Georgia Tech.
The project will mine data to rapidly understand interactions at what Prasanna calls “the edge of the internet.” Computing and graph processing will move to the “edge” near the source of the data and enable real-time decision-making as the data arrives without relaying back to the data centers. The software will be the critical component of the 1000-fold speed-up envisioned in the program, said Prasanna.
Although driven by national security goals, these technologies, said Kannan, will deliver practical benefits to consumers through “social media analysis to reduce traffic congestion, better product matching for online shopping, self-driving cars, and lower-cost electricity.”
“Extremely high-speed graph analytics streamed through the processor will enable consumers to have a more enjoyable experience at LA theme parks with less waiting and allow faster domain-specific “vertical” internet search,” said Kannan.