Dr. Khan is a Professor of Electrical and Computer Engineering at the University of Connecticut. He leads the Computer Architecture Group (CAG) and also serves as an Associate Director of the Connecticut Advanced Computing Center (CACC). His research interests include computer architectures that exploit parallelism and locality for high-performance applications, such as graph intelligence problems. He has contributed architectural advancements for futuristic massively parallel microprocessors that substantially enhance system level performance and efficiency. Prior to joining UConn, Khan was a Postdoctoral Research Scientist at the Massachusetts Institute of Technology. He received Ph.D. from the University of Massachusetts Amherst. He spent several years in the semiconductor industry as a high-performance processor architect.
Dr. Khan is addressing the computational complexity problem in artificial intelligence applications, such as autonomous systems, social influence, and chip design that must handle increasingly large and sparse graph-based data. Efficient processing of sparse graph problems is extremely challenging since the underlying computations require complex mathematical operations whose processing suffers from performance scaling challenges on existing hardware processing units. He is developing parallel architectures that exploit sparsity for performance to reduce computational complexity in graph intelligence problems.
As a complementary research effort, Dr. Khan is developing a system architecture that takes a hardware-architecture-algorithm approach to optimize multiple goals at once, like finding the best trade-off between speed and fuel efficiency for autonomous vehicles. His approach exploits massive parallelism and architectural enhancements to rapidly accelerate the computationally hard multi-objective graph intelligence problems. This means exact solutions that used to take hours to generate can be found in seconds or less. This allows decision-makers to have access to real-time information, leading to better decision-making in high-impact application scenarios.
Recruiting PhD students to work on massively parallel computer architectures for graph intelligence problems. Contact me via email.
PhD in Electrical and Computer Engineering, 2009
University of Massachusetts Amherst
BSc in Electrical and Computer Engineering, 2000
Michigan State University
My research interests can be generalized to the field of Computer Architecture and Systems. My current research deals with hardware–software mechanisms for parallelism, security, and resiliency of future parallel computer architectures. I enjoy simulating and building architecture prototypes.
Many emerging applications comprise real-time automated processing, interpretation, and intelligent decisions using large volumes of input data, while simultaneously decreasing the time necessary to arrive at a decision. The objective of this research is to explore both hardware and software parallelism challenges holistically, characterize key bottlenecks, and explore architectural methods that improve performance, lower energy, and lower programmer effort.
Computer systems have recently seen a rise of malicious exploits on processor hardware. Virtualization technologies expose hardware resources, thus requiring strong isolation and obfuscation guarantees for security. Moreover, with the advent of confidential computing, the authenticity of sensitive code execution requires fast remote attestations. This research aims to address these challenges by devising methods to secure parallel processors, while meeting the efficiency and responsiveness expectations of the system.
The focus of this reseaerch is to build architectural mechanisms and protocols that exploit application through hardware layers to co-optimize processor resiliency and efficiency.
I am the principal investigator for the NSF REU Site on Trustable Embedded Systems Security. To learn more about the research projects and highlights from past summers, visit the REU site’s official webpage
.js-id-Demo