## Competitive Online Scheduling with Level of Service (2000)

### Cached

### Download Links

- [www.comp.nus.edu.sg]
- [www.comp.nus.edu.sg]
- [cs.nyu.edu]
- [www.cz3.nus.edu.sg]
- DBLP

### Other Repositories/Bibliography

Venue: | In Proc. 7th Annual International Computing and Combinatorics Conference |

Citations: | 11 - 3 self |

### BibTeX

@INPROCEEDINGS{Chang00competitiveonline,

author = {Ee-chien Chang and Chee Yap},

title = {Competitive Online Scheduling with Level of Service},

booktitle = {In Proc. 7th Annual International Computing and Combinatorics Conference},

year = {2000},

pages = {453--462},

publisher = {Springer}

}

### OpenURL

### Abstract

Motivated by an application in thinwire visualization, we study an abstract on-line scheduling problem where the size of each requested service can be scaled down by the scheduler. Thus our problem embodies a notion of "Level of Service" that is increasingly important in multimedia applications. We give two schedulers FirstFit and EndFit based on two simple heuristics, and generalize them into a class of greedy schedulers. We show that both FirstFit and EndFit are 2-competitive, and any greedy scheduler is 3-competitive. These bounds are shown to be tight.

### Citations

728 |
Amortized efficiency of list update and paging rules
- Sleator, Tarjan
- 1985
(Show Context)
Citation Context ...roperty. Note that there is considerable interest in similar Quality of Service (QoS) issues in multimedia research. We use the standard notion of “competitiveness” in the sense of Sleator and Tar=-=jan [9] t-=-o judge the quality of our online schedulers. A scheduler S produces a feasible schedule S(I) for each instance I of our scheduling problem. Each S(I) has an associated merit, where merit(S(I)) ≥ 0.... |

407 | Resource reservation protocol (RSVP) – version 1 functional specification - Braden, Zhang, et al. - 1997 |

393 | A scheduling model for reduced CPU energy
- Yao, Demers, et al.
- 1995
(Show Context)
Citation Context ...ill see, our problem is different from theirs in several ways. Woeginger [10] studied a problem that has several of the features of our problem. Other online interval packing problems can be found in =-=[11, 5]-=-. Outline. In the next section, we formulate our on-line scheduling problem and give two schedulers, FirstFit and EndFit. FirstFit is based on a heuristic which always serves the most important residu... |

138 |
Competitive algorithms for server problems
- Manasse, McGeoch, et al.
- 1990
(Show Context)
Citation Context ...:= sup I merit(opt(I)) merit(S(I)) . Thus, we want schedulers S with C(S) ≥ 1 as small as possible. The original paging problem studied by Sleator and Tarjan is a special case of the k-server proble=-=m [7]-=-. There is a fairly large literature on competitive algorithms (e.g., [1, 2, 8]). The class of problems most closely related to ours is the online interval packing problem for a single server, where a... |

86 | Online interval scheduling
- Lipton, Tomkins
- 1994
(Show Context)
Citation Context ...1, 2, 8]). The class of problems most closely related to ours is the online interval packing problem for a single server, where a schedule is a subset of non-overlapping intervals. Lipton and Tomkins =-=[6]-=- study a variant where the input intervals are sorted by their left endpoints. They give a randomized scheduler that is 2-competitive. As we will see, our problem is different from theirs in several w... |

70 |
Tarjan, Amortized eciency of list update and paging rules
- Sleator, E
- 1985
(Show Context)
Citation Context ...roperty. Note that there is considerable interest in similar Quality of Service (QoS) issues in multimedia research. We use the standard notion of "competitiveness" in the sense of Sleator a=-=nd Tarjan [9]-=- to judge the quality of our online schedulers. A scheduler S produces a feasible schedule S(I) for each instance I of our scheduling problem. Each S(I) has an associated merit, where merit(S(I)) # 0.... |

62 | Energy-efficient algorithms
- Albers
(Show Context)
Citation Context ... ≥ 1 as small as possible. The original paging problem studied by Sleator and Tarjan is a special case of the k-server problem [7]. There is a fairly large literature on competitive algorithms (e.g.=-=, [1, 2, 8]-=-). The class of problems most closely related to ours is the online interval packing problem for a single server, where a schedule is a subset of non-overlapping intervals. Lipton and Tomkins [6] stud... |

34 | Optimal on-line algorithms for single-machine scheduling
- Hoogeveen, Vestjens
- 1996
(Show Context)
Citation Context ...ill see, our problem is different from theirs in several ways. Woeginger [10] studied a problem that has several of the features of our problem. Other online interval packing problems can be found in =-=[11, 5]-=-. Outline. In the next section, we formulate our on-line scheduling problem and give two schedulers, FirstFit and EndFit. FirstFit is based on a heuristic which always serves the most important residu... |

30 |
On-line scheduling of jobs with fixed start and end times
- Woeginger
- 1994
(Show Context)
Citation Context ...t where the input intervals are sorted by their left endpoints. They give a randomized scheduler that is 2-competitive. As we will see, our problem is different from theirs in several ways. Woeginger =-=[10]-=- studied a problem that has several of the features of our problem. Other online interval packing problems can be found in [11, 5]. Outline. In the next section, we formulate our on-line scheduling pr... |

21 | Foveation techniques and scheduling issues in thinwirevisualization
- Chang
- 1998
(Show Context)
Citation Context ...e, and any greedy scheduler is 3-competitive. These bounds are shown to be tight. 1 Introduction We study an abstract on-line scheduling problem motivated by visualization across a “thinwire” netw=-=ork [4, 3]-=-. An example of such a visualization problem is a server-client model where the server and client are connected by a thinwire (that is, a bandwidthlimited connection such as the Internet), with the se... |

18 | Self-organizing data structures
- Albers, Westbrook
- 1998
(Show Context)
Citation Context ... ≥ 1 as small as possible. The original paging problem studied by Sleator and Tarjan is a special case of the k-server problem [7]. There is a fairly large literature on competitive algorithms (e.g.=-=, [1, 2, 8]-=-). The class of problems most closely related to ours is the online interval packing problem for a single server, where a schedule is a subset of non-overlapping intervals. Lipton and Tomkins [6] stud... |

18 | Realtime visualization of large images over a thinwire
- Chang, Yap, et al.
- 1997
(Show Context)
Citation Context ...e, and any greedy scheduler is 3-competitive. These bounds are shown to be tight. 1 Introduction We study an abstract on-line scheduling problem motivated by visualization across a “thinwire” netw=-=ork [4, 3]-=-. An example of such a visualization problem is a server-client model where the server and client are connected by a thinwire (that is, a bandwidthlimited connection such as the Internet), with the se... |

13 | Supporting Adaptive Flows in a Quality of Service Architecture”, Multimedia Systems Journal - Campbell, Coulson, et al. - 1996 |

13 | D.Hutchison, “Specifying QoS for Multimedia Communications within Distributed Programming Environments - Waddington - 1996 |

1 |
editors. On-Line Algorithms
- McGeoch, Sleator
- 1992
(Show Context)
Citation Context ... ≥ 1 as small as possible. The original paging problem studied by Sleator and Tarjan is a special case of the k-server problem [7]. There is a fairly large literature on competitive algorithms (e.g.=-=, [1, 2, 8]-=-). The class of problems most closely related to ours is the online interval packing problem for a single server, where a schedule is a subset of non-overlapping intervals. Lipton and Tomkins [6] stud... |

1 |
Private commuication
- Arkin
- 1997
(Show Context)
Citation Context ... As # tends to zero, the FirstFit schedule will be the one shown in Figure 1. 4 Example 2. Another scheduler, EndFit, always serves according to the canonical optimal schedule of the pending requests =-=[3]-=-. On arrival of a new request, EndFit stops its current service (that is, preempt), recomputes the canonical optimal schedule of the pending requests, and then continues serving according to the recom... |