@inproceedings{Thammawichai:2016:10.1109/CDC.2016.7798501,
author = {Thammawichai, M and Kerrigan, EC},
doi = {10.1109/CDC.2016.7798501},
publisher = {IEEE},
title = {Feedback Scheduling for Energy-Efficient Real-Time Homogeneous Multiprocessor Systems},
url = {http://dx.doi.org/10.1109/CDC.2016.7798501},
year = {2016}
}

Download

TY  - CPAPER
AB  - Real-time scheduling algorithms proposed in theliterature are often based on worst-case estimates of taskparameters and the performance of an open-loop scheme cantherefore be poor. To improve on such a situation, one caninstead apply a closed-loop scheme, where feedback is exploitedto dynamically adjust the system parameters at run-time. Wepropose an optimal control framework that takes advantageof feeding back information of finished tasks to solve a realtimemultiprocessor scheduling problem with uncertainty intask execution times, with the objective of minimizing thetotal energy consumption. Specifically, we propose a linearprogramming-based algorithm to solve a workload partitioningproblem and adopt McNaughton’s wrap around algorithmto find the task execution order. Simulation results for aPowerPC 405LP and an XScale processor illustrate that ourfeedback scheduling algorithm can result in an energy savingof approximately 40% compared to an open-loop method.
AU  - Thammawichai,M
AU  - Kerrigan,EC
DO  - 10.1109/CDC.2016.7798501
PB  - IEEE
PY  - 2016///
TI  - Feedback Scheduling for Energy-Efficient Real-Time Homogeneous Multiprocessor Systems
UR  - http://dx.doi.org/10.1109/CDC.2016.7798501
UR  - http://hdl.handle.net/10044/1/40567
ER  - 

Download