Fuzzy scheduling of real-time ensemble systems

This paper addresses the problem of resource scheduling in real-time ensemble systems. An ensemble system uses multiple simple computational models (called “experts”) to produce its outputs. Real system requirements of ensemble systems (e.g., size, weight, power and cost constraints) often lead to limited availability of computational resources required to support concurrent execution of all their experts. In practical systems, uncertainties in execution time and resource utilization complicate even further the scheduling of these experts. We propose a fuzzy-logic feedback-based resource scheduler (FuzzyFES) that can provide real-time execution of all relevant experts while minimizing the impact of limited resources and uncertainties on the system performance. FuzzyFES consists of a fuzzy-logic controller (FZ), a task utilization adaptor (TUA) and a real-time task scheduler (RTS) working harmoniously in a closed loop with an ensemble system to be scheduled. By considering the uncertainties that may be present in the systems and deployment environments, FZ determines the total allowable CPU utilization for the ensemble system. TUA then calculates the amount of resource utilization to be allocated to each expert not exceeding the total allowable utilization. The assigned utilization from TUA ensures that critical experts achieve their best performance while guaranteeing minimum execution time needed by others. RTS creates a real-time schedule for the experts to execute on multiple processors according to the allotted utilization. Our performance evaluation of a case-study ensemble system with limited resources demonstrates that FuzzyFES can schedule experts to produce outputs closely similar to those of the same system with sufficient resources, although the limited-resource system has up to 40% fewer resources. The results also confirm FuzzyFES's efficiency and show that execution-time imprecision and occasional fluctuation of resource availability can be tolerated by at least 45% more than when the experts are scheduled in an open-loop manner.

[1]  Geoffrey E. Hinton,et al.  Adaptive Mixtures of Local Experts , 1991, Neural Computation.

[2]  M. Araújo,et al.  BIOMOD – a platform for ensemble forecasting of species distributions , 2009 .

[3]  Chenyang Lu,et al.  Feedback utilization control in distributed real-time systems with end-to-end tasks , 2005, IEEE Transactions on Parallel and Distributed Systems.

[4]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[5]  Renato J. O. Figueiredo,et al.  A New Architecture for Deriving Dynamic Brain-Machine Interfaces , 2006, International Conference on Computational Science.

[6]  Xianbo He,et al.  A  New Adaptive Performance Feedback Control Scheduling Model Oriented to the Embedded Soft Real-Time Systems , 2008, 2008 International Conference on Embedded Software and Systems Symposia.

[7]  Prapaporn Rattanatamrong,et al.  Dynamic Scheduling of Real-Time Mixture-of-Experts Systems on Limited Resources , 2014, IEEE Transactions on Computers.

[8]  Chenyang Lu,et al.  DEUCON: Decentralized End-to-End Utilization Control for Distributed Real-Time Systems , 2007, IEEE Transactions on Parallel and Distributed Systems.

[9]  Bing Du,et al.  Embedded Robust Control Real-Time Scheduling , 2008, 2008 International Conference on Computer Science and Software Engineering.

[10]  Sehjeong Kim,et al.  A Real-Time Scheduler Design for a Class of Embedded Systems , 2008, IEEE/ASME Transactions on Mechatronics.

[11]  Lijuan Cao,et al.  Support vector machines experts for time series forecasting , 2003, Neurocomputing.

[12]  Renato Figueiredo,et al.  Cyber-Workstation for Computational Neuroscience , 2009, Front. Neuroeng..

[13]  Kaushal K. Shukla,et al.  Real-time task scheduling with fuzzy uncertainty in processing times and deadlines , 2008, Appl. Soft Comput..

[14]  Rami G. Melhem,et al.  Multiple-resource periodic scheduling problem: how much fairness is necessary? , 2003, RTSS 2003. 24th IEEE Real-Time Systems Symposium, 2003.

[15]  Prapaporn Rattanatamrong,et al.  Mode Transition for Online Scheduling of Adaptive Real-Time Systems on Multiprocessors , 2011, 2011 IEEE 17th International Conference on Embedded and Real-Time Computing Systems and Applications.

[16]  Xingshe Zhou,et al.  Adaptive resource management architecture for distributed real-time embedded systems , 2009, SAC '09.

[17]  Robert E. Schapire,et al.  The strength of weak learnability , 1990, Mach. Learn..

[18]  Lior Rokach,et al.  Pattern Classification Using Ensemble Methods , 2009, Series in Machine Perception and Artificial Intelligence.

[19]  C. A. Smith,et al.  THE ESTIMATION OF GENE FREQUENCIES IN A RANDOM‐MATING POPULATION , 1955, Annals of human genetics.

[20]  Jens Vygen,et al.  The Knapsack Problem , 2012 .

[21]  D M Wolpert,et al.  Multiple paired forward and inverse models for motor control , 1998, Neural Networks.

[22]  Prapaporn Rattanatamrong,et al.  Real-time scheduling of mixture-of-experts systems with limited resources , 2010, HSCC '10.

[23]  Lui Sha,et al.  Handling Execution Overruns in Hard Real-Time Control Systems , 2002, IEEE Trans. Computers.

[24]  Mihaela van der Schaar,et al.  Distributed online Big Data classification using context information , 2013, 2013 51st Annual Allerton Conference on Communication, Control, and Computing (Allerton).

[25]  Giuseppe Lipari,et al.  Elastic Scheduling for Flexible Workload Management , 2002, IEEE Trans. Computers.

[26]  Binoy Ravindran,et al.  T-L plane-based real-time scheduling for homogeneous multiprocessors , 2010, J. Parallel Distributed Comput..

[27]  Giorgio C. Buttazzo,et al.  Resource Reservation in Dynamic Real-Time Systems , 2004, Real-Time Systems.

[28]  Chuen-Chien Lee,et al.  Fuzzy logic in control systems: fuzzy logic controller. I , 1990, IEEE Trans. Syst. Man Cybern..