Semi-online scheduling on two uniform parallel machines with initial lookahead

Abstract

This work studies the problem of semi-online scheduling on two uniform parallel machines with speeds 1 and s≥2, respectively. We introduce a novel  concept of initial lookahead in which any deterministic online algorithm has the full knowledge of the first k jobs at the beginning, while the remaining jobs are released one-by-one in the online over-list mode. The objective of the considered problem is to minimize the makespan. We focus on the case where the first k jobs are of a total processing time not less than (s+1)Δ where Δ is the largest job length, and it is assumed that s is an integer. We prove a lower bound of (s2+s+1)/(s2+s), and propose a deterministic online algorithm with competitive ratio of (s+1)2/(s2+s+1). The ratio is at most 9/7 and much less than that of 1.618 for the basic case without initial lookahead (Cho and Sahni [7]).  Our results demonstrate that a finite ability of initial lookahead can greatly improve the competitiveness of online algorithms.