Scheduling data transfers in parallel computers and communications systems
Scheduling data transfers in parallel computers and communications systems
On-line scheduling of jobs with fixed start and end times
Theoretical Computer Science - Special issue on dynamic and on-line algorithms
Heuristics for Scheduling I/O Operations
IEEE Transactions on Parallel and Distributed Systems
Scheduling time-constrained communication in linear networks
Proceedings of the tenth annual ACM symposium on Parallel algorithms and architectures
Time-constrained scheduling of weighted packets on trees and meshes
Proceedings of the eleventh annual ACM symposium on Parallel algorithms and architectures
Simple competitive request scheduling strategies
Proceedings of the eleventh annual ACM symposium on Parallel algorithms and architectures
Scheduling with conflicts, and applications to traffic signal control
Proceedings of the seventh annual ACM-SIAM symposium on Discrete algorithms
SODA '94 Proceedings of the fifth annual ACM-SIAM symposium on Discrete algorithms
Patience is a virtue: the effect of slack on competitiveness for admission control
Proceedings of the tenth annual ACM-SIAM symposium on Discrete algorithms
Developments from a June 1996 seminar on Online algorithms: the state of the art
Hi-index | 0.00 |
We consider the online competitiveness for scheduling a set of communication jobs (best described in terms of a weighted graph where nodes denote the communication agents and edges denote communication jobs and three weights associated with each edge denote its length, release time, and deadline, respectively), where each node can only send or receive one message at a time. A job is accepted if it is scheduled without interruption in the time interval corresponding to its length between release time and deadline. We want to maximize the sum of the length of the accepted jobs. When an algorithm is not able to preempt (i.e., abort) jobs in service in order to make room for better jobs, previous lower bound shows that no algorithm can guarantee any constant competitive ratio. We examine a natural variant in which jobs can be aborted and each aborted job can be rescheduled from start (called restart). We present simple algorithms under the assumptions on job length: 2-competitive algorithm for unit jobs under the discrete model of time and (6+4√2 ≈ 11:656)-competitive algorithm for jobs of arbitrary length. These upper bounds are compensated by the lower bounds 1.5, 8-Ɛ, respectively.