Keywords and Phrases
Directed Acyclic Graphs (DAG); Energy Efficiency; Heterogeneous Processors; ODROID; Real-Time System; RTOS
"Increasing number of battery operated devices creates a need for energy-efficient real-time operating system for such devices. Designing a truly energy-efficient system is a multi-staged effort; this thesis consists of three main tasks that address different aspects of energy efficiency of a real-time system (RTS).
The first chapter introduces an energy-efficient algorithm that alternates processor frequency using DVFS to schedule tasks on cores. Speed profiles is calculated for every task that gives information about how long a task would run for and at what processor speed. We pair tasks with similar speed profiles to give us a resultant merged speed profile that can be efficient scheduled on a cluster. Experiments carried out on ODROID-XU3 are compared with a reference approach that provides energy saving of up to 20%.
The second chapter proposes power-aware techniques to segregate a task set over a heterogeneous platform such that the overall energy consumption is minimized. With the help of calculated speed profiles, second contribution of this work feasibly partitions a given task set into individual sets for a cluster based homogeneous platform. Various heuristics are proposed that are compared against a baseline approach with simulation results.
The final chapter of this thesis focuses on the importance of having an underlying energy-efficient operating system. We discuss an energy-efficient way of porting a real-time operating system (RTOS), QP, over TMS320F28377S along with modifications to make the Operating System (OS) consume minimal energy for its operation"--Abstract, page iii.
Zawodniok, Maciej Jan, 1975-
Kimball, Jonathan W.
Electrical and Computer Engineering
M.S. in Computer Engineering
Missouri University of Science and Technology
xi, 51 pages
© 2018 Aamir Aarif Khan, All rights reserved.
Thesis - Open Access
Electronic OCLC #
Khan, Aamir Aarif, "Developing an energy efficient real-time system" (2018). Masters Theses. 7799.