1. Real-Time Systems Laboratory @ Hierarchical Real-Time Systems for Imprecise Computation Model The 5th EuroSys Doctoral Workshop (EuroDW 2011) Guy Martin Tchamgoue Department of Informatics, Gyeongsang National University, South Korea

2. Hierarchical Real-Time Systems for Imprecise Computation Model  Problem  An Application  Proposed Solution  Preliminary Results  Ongoing Work EuroDW'11 April 10, 2011 2 Agenda

3. Hierarchical Real-Time Systems for Imprecise Computation Model  Statement  To build large complex real-time systems from simpler ones  Flexibility and Reliability  To handle both hard and soft real-time systems  To enhance the resource utilization on system overload  To facilitate the development and deployment process EuroDW'11 April 10, 2011 3 Problem

4. Hierarchical Real-Time Systems for Imprecise Computation Model EuroDW'11 April 10, 2011 4 An Application Operating System

5. Hierarchical Real-Time Systems for Imprecise Computation Model  Hierarchical resource sharing among components under different scheduling policies  Higher-level schedulers :  Consider each component as a scheduling unit  Do not need to know about the internal complexity of its components  Lower-level schedulers :  Focus on local scheduling for a given resource supply EuroDW'11 April 10, 2011 5 Compositional Scheduling Framework C1C1 C2C2 C3C3 C4C4 C1C1 C2C2

6. Hierarchical Real-Time Systems for Imprecise Computation Model  Publications  Tchamgoue et al., “Compositional Scheduling Framework for Imprecise Computation Model”, Work-in-Progress Session, RTAS’10, April 2010  Tchamgoue et al., “Hierarchical Real-Time Scheduling Framework for Imprecise Computations”, 8 th IEEE/IFIP EUC, December 2010  System Model  Imprecise Computation Tasks: T = {  i (p i, m i, o i, f i )}  p i : period  m i : execution time of the mandatory part  o i : execution time of the optional part  f i : reward function for optional execution EuroDW'11 April 10, 2011 6 Preliminary Results

7. Hierarchical Real-Time Systems for Imprecise Computation Model  The Imprecise Resource Model :  ( , ,  )   : resource period   : resource allocation for mandatory part   : resource allocation for optional part   ( , ,  ) periodically supplies at least  time units and at most  +  time units.  Example:  The resource supply of  ( , ,  ) EuroDW'11 April 10, 2011 7

8. Hierarchical Real-Time Systems for Imprecise Computation Model  The Interface Model  Specifying the real-time requirement of a component  Interface Model I(P, M, O, F)  P : period  M : mandatory part requirement  O : optional part requirement  F : reward function of optional execution EuroDW'11 April 10, 2011 8  1 (40, 5, 3, f 1 (t))  2 (30, 4, 2, f 2 (t)) MF-EDF I 1 (10, 3, 2, F 1 (t))  3 (50, 4, 2, f 3 (t))  4 (25, 3, 3, f 4 (t)) MF-RM I 2 (15, 4, 2, F 2 (t))  5 (10, 3, 2, F 1 (t))  6 (15, 4, 2, F 2 (t)) MF-EDF I 0 (5, 3, 2, F 0 (t)) Component C 0 Component C 1 Component C 2 R 1 =  1 (10, 3, 2) R 2 =  2 (15, 4, 2)

9. Hierarchical Real-Time Systems for Imprecise Computation Model  Other Results  Schedulability Analysis  To guarantee the minimum requirements of a component  Utilization Bound  To determine the largest possible utilization bound that makes a component schedulable  Scheduling Algorithm with Guarantee of Reward  To provide a minimum reward guarantee to an upper layer that provides resources to lower layers EuroDW'11 April 10, 2011 9

10. Hierarchical Real-Time Systems for Imprecise Computation Model  Component Interface Generation  How to optimally compute the resource and interface parameters  Component Reward function  To derive the reward function for the component’s interface  Optimal scheduling of optional parts  To schedule optional parts in order to maximize the total reward of a component EuroDW'11 April 10, 2011 10 Ongoing Work

11. Hierarchical Real-Time Systems for Imprecise Computation Model EuroDW'11 April 10, 2011 11 Thank you for your attention. EuroDW 2011

12. Hierarchical Real-Time Systems for Imprecise Computation Model  Proposed for flexible scheduling of hard deadline tasks  An IC task consists of:  A mandatory subtask for an acceptable result, and  An optional subtask for  refining the mandatory task result,and  improving the quality of the task according to the amount of execution time. EuroDW'11 April 10, 2011 12 Imprecise Computation* (IC) Model * J. W.-S. Liu, K.-J. Lin, W.-K. Shih, A. C.-S. Yu, C. Chung, J. Yao, and W. Zhao, “Algorithms for scheduling imprecise computations,” IEEE Computer, vol. 24, no. 5, pp. 58–68, 1991. period time : Mandatory part: Optional part