An On-Line Scheduling Heuristic With Better Worst Case Ratio Than Graham's List Scheduling

G. Galambos,G. Woeginger

DOI: https://doi.org/10.1137/0222026

1993-04-01

Abstract:The problem of on-line scheduling a set of independent jobs on m machines is considered. The goal is to minimize the makespan of the schedule. Graham’s List Scheduling heuristic [R. L. Graham, SIAM J. Appl. Math., 17(1969), pp. 416–429] guarantees a worst case performance of $2 - \frac{1} {m}$ for this problem. This worst case bound cannot be improved for $m = 2$ and $m = 3$. For $m \geqslant 4$, approximation algorithms with worst case performance at most $2 - \frac{1}{m} - \varepsilon _m $ are presented, where $\varepsilon _m $ is some positive real depending only on m.