Research Area

Standby-sparing systems

Multiprocessor systems in which, each CPU core has an associated backup or spare core. When a software or hardware fault occurs, the spare CPU can take over and re-run the task, producing fault-free results and maintaining the deadline of the original task.

Parallel and distributed computing

Computing systems made up of multiple parallel processing modules, interconnected through high-speed network. Major challenges for these type of systems are minimizing energy consumption, reliability of the network, and overall system throughput.

Safety-critical design

Dependable computing. Systems that have multiple levels of fail-safe mechanisms to deal with transient, permanent, or intermittent faults. Extensively used in space-missions and autonomous driving technologies, they provide hard timing guarantees to the application-tasks even in the face of fault-bursts.

Multicore embedded systems

Control-systems built on chip-multiprocessors, specifically, heterogeneous multicore processors (HMPs). The multiple types of cores can have distinctively different power and performance characteristic while supporting a single instruction set architecture. Minimizing energy consumption for a range of different types of workload is a major challenge here.

Cyber physical systems

Most popular class of systems today, and covers a wide range of devices and applications. Exploring new design possibilities, improving the reliability and efficiency of tightly coupled subsystems, minimizing energy consumption.

Discreet event simulator

Many of the systems under consideration are highly conceptual and futuristic, meaning they are yet to be built practically. So we built a discreet event simulator that can mimic power and timing characteristics of big.Little systems executing periodic tasks. We continuously put effort in adding more features to it and thereby making it reflect reality more accurately.

Publications

IGSC 2020

Abhishek Roy, Hakan Aydin, and Dakai Zhu. “Energy-Aware Primary/Backup Scheduling of Periodic Real-Time Tasks on Heterogeneous Multicore Systems.” Proceedings of the 11th International Green and Sustainable Computing Conference (IGSC), IEEE, October, 2020. (Best paper award) View | Download

IGSC 2019

Abhishek Roy, Hakan Aydin, and Dakai Zhu. “Energy-Efficient Fault Tolerance for Real-Time Tasks with Precedence Constraints on Heterogeneous Multicore Systems.” Proceedings of the 10th International Green and Sustainable Computing Conference (IGSC), IEEE, Alexandria, Virginia, October, 2019. View | Download

IGSC 2017

Abhishek Roy, Hakan Aydin, and Dakai Zhu. “Energy-Efficient Primary/Backup Scheduling Techniques for Heterogeneous Multicore Systems.” Proceedings of the 8th International Green and Sustainable Computing Conference (IGSC), IEEE, Orlando, Florida, Oct 23-25, 2017. View | Download

DAC 2017

Abhishek Roy, Hakan Aydin, and Dakai Zhu. “Energy-Aware Standby-Sparing on Heterogeneous Multicore Systems.” Proceedings of the 54th Annual Design Automation Conference (DAC), ACM, 2017. View | Download

RTNS 2016

Abhishek Roy, Hakan Aydin, and Dakai Zhu. “On Task Period Assignment in Multiprocessor Real-Time Control Systems.” Proceedings of the 24th International Conference on Real-Time Networks and Systems (RTNS), ACM, 2016. Brest, France. View | Download

Education

Doctor of Philosophy in Computer Science

George Mason University, Fairfax, Virginia.
May 2021
Dissertation Title: Improving Energy Efficiency and Quality-of-Control Metrics in Reliable Multiprocessor Real-Time Systems.

Master of Science in Computer Science

George Mason University, Fairfax, Virginia.
May 2018

Bachelor of Science in Computer Science and Engineering

Bangladesh University of Engineering and Technology (BUET), Dhaka, Bangladesh.
February 2009.

Work Experience

Graduate Research Assistant
Dept. of Computer Science, George Mason University
May 2014 - Present
  • Conducting research on real-time multi-processor and multi-core systems with a goal to minimize energy consumption while maintaining fault tolerance.
  • Working on improving quality-of-control of real-time digital controllers implemented on multiprocessor platforms by optimizing resource allocation.
  • Supported by an NSF CSR grant.
Graduate Lecturer (Instructor on record)
Dept. of Computer Science, George Mason University
August 2012 - May 2014
  • Instructed CS-262: Intro. to Low-Level Programming in C (90+ students)
  • Supervised several Teaching Assistants for CS-262
Graduate Teaching Assistant
Dept. of Computer Science, George Mason University
August 2012 - May 2014
  • Assisted instructors in preparing assignments and grading homeworks for courses including Operating Systems, Databases, and Computer Networks.
Software Engineer
Nilavo Technologies Limited, Dhaka, Bangladesh
April 2010 - August 2011
  • Web application:- Prototyped a web application for company equity management as a new product, and primarily developed it using Agile process. ASP.NET, C# and MS SQL Server.
  • Software development:- Followed process based development cycles (agile, scrum).
  • Helped team for bug fixing in a file delivery application developed with a Java EE web framework.
System Analyst
Cashlink Bangladesh Limited, Dhaka, Bangladesh
March 2009 - April 2010
  • Transaction data:- Maintained a network of Automatic Teller Machines, which are supported by a number of commercial banks (clients).
  • Big data systems:- Issued queries to the transaction switch(IBM AS400 midrange with DB2) in order to settle disputes and facilitate maintenance of ATMs.
  • Security:- Familiar with secure working environment (3DES, HSM and VPN).
  • SQL and Perl:- Developed complex database queries for transaction monitoring and report generation.
  • Hardware:- Prototyped the first point of sales (POS) devices deployed in Bangladesh having EMV chip cards.
  • Data mirroring:- Mirrored part of the DB2 database to an MS SQL Server in order to provide the client banks direct access to the Card Management System without overloading the production servers.