Microelectronics has been at the center of a national conversation, in large part as a result of the COVID-19 pandemic. A global chip shortage amid a surge in coronavirus infections has forced the US to grapple with the downside of semiconductor manufacturing offshoring.
The issue has highlighted the importance of research projects such as those at the Secure, Trusted and Assured Microelectronics, or STAM Center, led by Michel Kinsy, an associate professor of computer science and engineering at the School of Computing and Augmented Intelligence at the Ira A. Fulton Schools of Engineering from Arizona State University.
Two directors of the STAM center, computer science PhD students Mihailo Isakov and Alan Ehrlich are working with Kinsy to create a foundation for future secure and trusted semiconductor and microelectronics technologies.
“Alan and I have Dr. Met Kinsy while pursuing PhDs at other universities; We have followed him across several institutions. dr Kinsy is an incredible leader with an amazing commitment to his students, so following him here wasn’t even a question,” says Isakov. “There was a lot of synergy for us coming to Arizona because of the access to semiconductor manufacturing and security experts. It’s probably the best place in the US for our research.”
The STAM Center has enabled Isakov to focus on machine learning, artificial intelligence, deep learning, hardware accelerators, and introspection and self-aware systems as research director for the Artificial Intelligence Technology and Systems, or AITS, Laboratory.
With encouragement from Kinsy, Isakov and Ehrlich are leading several exciting research projects, with Isakov’s work ranging from reinterpreting the way supercomputers exchange information and collaborate with other machines to collaborating with MIT’s Lincoln Laboratory on radar image analysis and secure machine learning systems .
“There is a certain joy in working with hardware; it has real staying power,” says Isakov. “It’s amazing to be able to close with a tangible end result like a chip. I also have a love and appreciation for software, but working with hardware has been amazing.”
Last summer, Isakov was made a WJ Cody Associate at Argonne National Laboratory, where he helped design particle accelerators and detectors for the European Organization for Nuclear Research, known as CERN.
Additionally, his recent research paper, A Taxonomy of Error Sources in HPC I/O Machine Learning Model – which pinpoints where supercomputers are underutilized and addresses problem areas – has been accepted for publication at this year’s SC22 International Conference for High Performance Computing, Networking, Storage and Analysis, the premier conference in high performance computing.
“DR. Kinsy urges us all to explore as many areas of engineering as possible in order to become scientists, not just computer scientists,” says Isakov isn’t in his area of expertise. The mentality that we have at STAM Center is that you can learn anything if you put yourselves into it. It’s a very powerful feeling because now I have the skills to push my way into areas that I’m in I would never have seen myself.”
Among their many research projects, Isakov and Ehrlicht also contribute to a center-wide effort called the Trireme. Ehrlich’s contributions to Trireme are among his proudest achievements in his doctoral experience.
“Together with the other members of the STAM Center, I’ve been working on the latest version of Trireme for more than five years,” says Ehrlich. “It’s an internal hardware platform, so everything starts in the STAM Center, which needs a processor in it, pretty much here. We use it as a platform to explore the design space, so in many ways it’s the toolbox for all of our architectural research.”
Trireme, a set of RISC-V based processors and memory hierarchies, is the infrastructure at the heart of many of STAM Center’s research projects.
Because RISC-V is an open and royalty-free instruction set architecture, or ISA, Trireme offers a high level of customization and an attractive option for many applications ranging from machine learning accelerators to secure computing system design.
The resource allows students and researchers to experiment with RISC-V ISA features and quickly complete functional architectures, making it a highly desirable tool to present at research conferences across the country.
“I was just able to trireme with Dr. Kinsy in Washington, DC at the Tapia 2022 Conference, which celebrates diversity and inclusion in computing,” says Ehrlich. “It has been so rewarding to see how this platform has grown over the years. Now it’s in a place where we can offer it as a resource and provide tutorials for other universities who will use it for teaching and as a platform for their own research.”
At the STAM Center, Ehrlich is the Research Director for the Computer Architecture and Embedded Systems Laboratory, or CAES, where he focuses on hardware security and building everything needed for the Center’s projects. As a senior researcher, he also has a mentoring role, guiding other students through their own research and helping them decipher projects he has devoted years to understanding.
Ehrlich’s other research projects focus on ultra-low-power hardware and secure data center-scale storage systems. In his thesis, he examines how to build a complete computing continuum by connecting edge and cloud systems such as a credit card reader and a data center to process transactions. His work examines how these systems can be unified to support secure data exchange and protect information such as credit card numbers everywhere from the checkout to the data center in the cloud.
“Besides our research, the STAM Center’s human resource development mission is something I’m proud of,” says Ehrlich. “We are constantly recruiting motivated PhD students interested in cybersecurity. We are constantly networking with students to show the opportunities and possibilities out there. I look forward to continuing our outreach work at the STAM Center after graduating as a Principal Investigator for the CAES lab, where I will also help lead the deployment of our current chip design, fabrication and test efforts .”
The center’s future looks bright thanks to the CHIPS and Science Act of 2022 — passed in part as a result of challenges posed by chip shortages in the US — which will boost semiconductor manufacturing here.
“We’ve dedicated years of our lives to this research, so it’s very comforting to know that Congress is agreeing that this is a national priority,” says Ehrlich. “It is incredible to see that our center’s mission aligns with what is close to the heart of our nation’s leaders. That’s a great feeling as a researcher. We are excited to see what the future will bring.”