Shareef Ahmed is currently a Ph.D. student in the Department of Computer Science at the University of North Carolina at Chapel Hill, working with Prof. Jim Anderson. His research interests span across real-time systems, distributed and concurrent algorithms, and resource allocation algorithms. His current research focuses on providing tight analysis for different scheduling and synchronization algorithms. His work has appeared in top real-time systems conferences including RTSS, RTAS, and ECRTS.
Recent advances in artificial-intelligence (AI) techniques have fueled an increased demand for autonomous features in real-time systems. A system with such features usually involves both sequential and parallel real-time tasks with data-flow dependencies and requires multicore platforms along with hardware accelerators (e.g., graphics processing units) for efficient computation. Unfortunately, how to best schedule such systems while guaranteeing real-time constraints has not been thoroughly investigated. The development of such scheduling techniques is also hampered by prevalent pessimism in schedulability analysis of simpler systems, e.g., systems with only independent sequential tasks. My work addresses these issues by providing efficient algorithms to compute tight response-time bounds for different complex task models under common real-time scheduling techniques. Additionally, my work enables the design of truly (not just asymptotically) optimal real-time synchronization protocols.