A Heuristic Scheduling Algorithm for Reducing the Total Error of an Imprecise Multiprocessor System with 0/1 Constraint

  • Published : 1997.12.01

Abstract

The scheduling problem of satisfying both 0/1 constraint and the timing constraint while minimizing the total error is NP-complete when the optional parts have arbitrary processing times. In this paper, we present a heuristic scheduling algorithm for 0/1 constraint imprecise systems which consist of communicating tasks running on multiple processors. The algorithm is based on the program graph which is similar to the one presented in[4]. To check the schedulability, we apply Lawler and Moore's theorem. To analyze the performance of the proposed algorithm, intensive simulation is done. The results of the simulation shows that the longest processing first selection strategy outperforms random or minimal laxity policies.

Keywords

References

  1. Ph.D. Thesis, University of Cambridge Operating Support for Quality of Services E.A.Hyden
  2. In Foundation of Real-Time Computing, Scheduling and Resource Management Algorithms for Scheduling Imprecise Computations Liu et al
  3. Management Science v.12 Scheduling with deadlines and Loss Functions R.McNaughton
  4. Proc. IEEE Real-Time System Symposium Dynamic End-to-End Guarantees in Distributed Real Time Systems M.C.Natale;J.Stankovic
  5. Computer v.24 no.5 Building Flexible Real-Time Systems Using the FLEX Language S.Natarajan,Kenny;K.J.Lin
  6. Proc. IEEE Real-Time Systems Symposium Program Representation and Translation to Support Predictability of Real-Time Systems D.Niehaus
  7. Proc. IEEE Real-Time Systems Symposium Fast Algorithms for Scheduling Tasks with Ready Times and Deadlines to Minimize Total Error W.K.Shih;J.W.Liu et al