### Abstract

Every task in a real-time system has a deadline by which time it should complete and a vaJue that it obtains only if it completes by its deadline. The problem is to design an on-line scheduling algorithm (i.e., the scheduler has no knowledge of a task until it is released) that maximizes the obtained value. When such a system is underloaded (i.e. there exists a schedule for which all tasks meet their deadlines), Dertouzos showed that the earliest deadline first algorithm will achieve 100% of the possible value. Locke showed that earliest deadline first performs very badly when the system is overloaded and proposed heuristics to deal with overload. Baruah et al. showed that the best that an on-line algorithm can guarantee is 1/(1+k)^{2} of the value that an off-line (clairvoyant) algorithm can obtain where k is the ratio between the highest value density and the lowest value density task in the system (where the value density of a task is its value divided by its computation time). We present an on-line algorithm that achieves the above best possible guarantee in this model. This algorithm also obtains 100% of the possible value when the system is underloaded.

Original language | English (US) |
---|---|

Title of host publication | Proceedings - Real-Time Systems Symposium, RTSS 1992 |

Pages | 290-299 |

Number of pages | 10 |

DOIs | |

State | Published - 1992 |

Event | 1992 Real-Time Systems Symposium, RTSS 1992 - Phoenix, AZ, United States Duration: Dec 2 1992 → Dec 4 1992 |

### Other

Other | 1992 Real-Time Systems Symposium, RTSS 1992 |
---|---|

Country | United States |

City | Phoenix, AZ |

Period | 12/2/92 → 12/4/92 |

### Fingerprint

### ASJC Scopus subject areas

- Computer Networks and Communications
- Hardware and Architecture
- Software

### Cite this

*Proceedings - Real-Time Systems Symposium, RTSS 1992*(pp. 290-299). [242650] https://doi.org/10.1109/REAL.1992.242650

**Dover : An optimal on-line scheduling algorithm for overloaded real-time systems.** / Koren, Gilad; Shasha, Dennis.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Proceedings - Real-Time Systems Symposium, RTSS 1992.*, 242650, pp. 290-299, 1992 Real-Time Systems Symposium, RTSS 1992, Phoenix, AZ, United States, 12/2/92. https://doi.org/10.1109/REAL.1992.242650

}

TY - GEN

T1 - Dover

T2 - An optimal on-line scheduling algorithm for overloaded real-time systems

AU - Koren, Gilad

AU - Shasha, Dennis

PY - 1992

Y1 - 1992

N2 - Every task in a real-time system has a deadline by which time it should complete and a vaJue that it obtains only if it completes by its deadline. The problem is to design an on-line scheduling algorithm (i.e., the scheduler has no knowledge of a task until it is released) that maximizes the obtained value. When such a system is underloaded (i.e. there exists a schedule for which all tasks meet their deadlines), Dertouzos showed that the earliest deadline first algorithm will achieve 100% of the possible value. Locke showed that earliest deadline first performs very badly when the system is overloaded and proposed heuristics to deal with overload. Baruah et al. showed that the best that an on-line algorithm can guarantee is 1/(1+k)2 of the value that an off-line (clairvoyant) algorithm can obtain where k is the ratio between the highest value density and the lowest value density task in the system (where the value density of a task is its value divided by its computation time). We present an on-line algorithm that achieves the above best possible guarantee in this model. This algorithm also obtains 100% of the possible value when the system is underloaded.

AB - Every task in a real-time system has a deadline by which time it should complete and a vaJue that it obtains only if it completes by its deadline. The problem is to design an on-line scheduling algorithm (i.e., the scheduler has no knowledge of a task until it is released) that maximizes the obtained value. When such a system is underloaded (i.e. there exists a schedule for which all tasks meet their deadlines), Dertouzos showed that the earliest deadline first algorithm will achieve 100% of the possible value. Locke showed that earliest deadline first performs very badly when the system is overloaded and proposed heuristics to deal with overload. Baruah et al. showed that the best that an on-line algorithm can guarantee is 1/(1+k)2 of the value that an off-line (clairvoyant) algorithm can obtain where k is the ratio between the highest value density and the lowest value density task in the system (where the value density of a task is its value divided by its computation time). We present an on-line algorithm that achieves the above best possible guarantee in this model. This algorithm also obtains 100% of the possible value when the system is underloaded.

UR - http://www.scopus.com/inward/record.url?scp=84880867136&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84880867136&partnerID=8YFLogxK

U2 - 10.1109/REAL.1992.242650

DO - 10.1109/REAL.1992.242650

M3 - Conference contribution

AN - SCOPUS:84880867136

SN - 0818631953

SN - 9780818631955

SP - 290

EP - 299

BT - Proceedings - Real-Time Systems Symposium, RTSS 1992

ER -