M. Hassan Najafi has been awarded the European Design and Automation Association’s 2018 Outstanding Dissertations Award for his research contributions to the advancement of design, automation, and test. Established to promote the work of young researchers in the field, and to recognize the importance of university research, the award is open to researchers across the world. Hassan, a recent graduate of ECE’s doctoral program, is one of four award winners. The award will be conferred at the Design, Automation and Test in Europe 2019 Conference and Exhibition to be held in Florence, Italy, during the last week of March.
Hassan’s dissertation is titled “New Views for Stochastic Computing: From Time-Encoding to Deterministic Processing.” Stochastic computing first emerged in the 1960s, positioned as a new paradigm for emerging technologies and post-CMOS computing, and an alternative to conventional computing. While there are several advantages to stochastic computing, latency of operations, which translates to higher energy consumption, is a noteworthy disadvantage. The cost of bit-stream generators is another important downside.
In his dissertation, Hassan addresses these issues by proposing an unorthodox idea: performing computation with digital constructs on time-encoded analog signals. He presents a new, energy-efficient, high-performance and inexpensive approach for stochastic computing using time-encoded pulse signals. His experimental results have shown several gains: 99% performance speedup, 98% savings in energy loss, and 40% area reduction compared to prior stochastic implementations.
Hassan also addresses achieving progressive precision, another key challenge with deterministic methods of stochastic computing, He proposes a high-quality down sampling method which significantly improves processing time and energy consumption of typical deterministic methods. His research offers two noteworthy firsts: introduction of two novel deterministic methods of processing bit-streams using low-discrepancy sequences, and exploitation of the skew tolerance of stochastic circuits to develop polysynchronous clocking. Most significantly, as an outcome of his research, Hassan presents a seamless stochastic system, StochMem, which uses analog memory to trade the energy and area overhead of data conversion for computation accuracy in stochastic systems.
IMPACT OF HASSAN’S RESEARCH
In the course of his research, Hassan developed new design methodologies, and new research directions to the stochastic computing and unary processing fields. His work has established some counterintuitive and fundamental new design methodologies for the design of digital stochastic systems. The results of his research have challenged the perceived limitations of stochastic computing, while simultaneously paving a path to designing significantly smaller, faster, and energy-efficient embedded systems.
AWARDS AND HONORS
Hassan has been a recipient of several awards and honors during his time as a student in the Department of Electrical and Computer Engineering. Most recently, he received the University’s 2017-2018 Doctoral Dissertation Fellowship Award. The dissertation fellowship is a University-wide award that supports top doctoral candidates.
His work on polysynchronous stochastic circuits was selected as the Feature Paper of the Month in the October 2017 issue of IEEE Transactions on Computers. His work on high-quality down-sampling of deterministic bit-streams was presented in the Best Papers Session of the 35th IEEE International Conference on Computer Design (ICCD 2017) and received the Best Paper Award of the conference. The paper was also among the top-ranked papers of the conference for publication in IEEE Transactions on Emerging Topics in Computing: Special Issue on Emerging Topics in Computer Design.
M. Hassan Najafi earned his doctoral degree in electrical engineering at the University of Minnesota in 2018, under the guidance of Prof. David Lilja. Currently, he is an Assistant Professor at the Center for Advanced Computer Studies at the University of Louisiana at Lafayette. Learn more about Hassan’s research here.