Speed is as powerful as clairvoyance
Journal of the ACM (JACM)
Buffer overflow management in QoS switches
STOC '01 Proceedings of the thirty-third annual ACM symposium on Theory of computing
An optimal online algorithm for packet scheduling with agreeable deadlines
SODA '05 Proceedings of the sixteenth annual ACM-SIAM symposium on Discrete algorithms
Better online buffer management
SODA '07 Proceedings of the eighteenth annual ACM-SIAM symposium on Discrete algorithms
Considering suppressed packets improves buffer management in QoS switches
SODA '07 Proceedings of the eighteenth annual ACM-SIAM symposium on Discrete algorithms
Optimal buffer management via resource augmentation
ISAAC'04 Proceedings of the 15th international conference on Algorithms and Computation
Online scheduling of packets with agreeable deadlines
ACM Transactions on Algorithms (TALG)
| Hi-index | 0.00 |
We consider the following scheduling problem: Given a set of unit-length jobs characterized by release times, deadlines and weights an algorithm must pick at most one job for each unit of time with the goal of maximizing the total weight of the chosen jobs. An on-line algorithm is an algorithm which has no information about a job before its release time. In the basic version of this problem it is known that any on-line algorithm is outperformed by an optimal off-line algorithm by a factor of ${{1+\sqrt{5}}\over{2}}-\varepsilon$ on some instance. In this paper we examine the effect of increasing the on-line algorithm's speed --- that is the number of jobs chosen for each unit of time --- and show that no speedup can fully compensate for the lack of complete information.