An Exact Algorithm for Non-preemptive Peak Demand Job Scheduling

Yaw, Sean; Mumey, Brendan

doi:10.1007/978-3-319-12691-3_1

Cited by 6 publications

(9 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our work introduces the first general purpose approximation algorithm where no assumptions are placed on the job parameters. Finally, optimal algorithms have been introduced that work on a limited number of jobs [3,8]. Our work also introduces an optimal fixed-parameter tractable algorithm that is able to schedule approximately 15 times more jobs in practice than previous approaches.…”

Section: Related Workmentioning

confidence: 99%

“…Peaks in power demand are proportionally more expensive to generate and provision for (since more infrastructure is required), so it is advantageous to schedule power-consuming jobs in such a way as to minimize peak demand. This problem has been previously formalized as the Peak demand minimization (PDM) problem, and has been studied extensively [1][2][3][4][5][6][7][8]. The basic formulation of the problem is as follows: Each job j is non-preemptible, meaning once it begins execution, it must run to completion without any interruptions.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Scheduling Non-Preemptible Jobs to Minimize Peak Demand

Yaw

Mumey

2017

Algorithms

Self Cite

View full text Add to dashboard Cite

This paper examines an important problem in smart grid energy scheduling; peaks in power demand are proportionally more expensive to generate and provision for. The issue is exacerbated in local microgrids that do not benefit from the aggregate smoothing experienced by large grids. Demand-side scheduling can reduce these peaks by taking advantage of the fact that there is often flexibility in job start times. We focus attention on the case where the jobs are non-preemptible, meaning once started, they run to completion. The associated optimization problem is called the peak demand minimization problem, and has been previously shown to be NP-hard. Our results include an optimal fixed-parameter tractable algorithm, a polynomial-time approximation algorithm, as well as an effective heuristic that can also be used in an online setting of the problem. Simulation results show that these methods can reduce peak demand by up to 50% versus on-demand scheduling for household power jobs.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Scheduling Non-Preemptible Jobs to Minimize Peak Demand

Yaw

Mumey

2017

Algorithms

Self Cite

View full text Add to dashboard Cite

show abstract

“…by (3). Since x i,j are independent Bernoulli trials, for any given t, with E[x i,j ] = p i,j and 0 < h j h ≤ 1 for all jobs, the following bound exists for all δ > 0: 2…”

Section: Theoremmentioning

confidence: 99%

“…Peaks in power demand are proportionally more expensive to generate and provision for (since more infrastructure is required), so it is advantageous to schedule power consuming jobs in such as way as to minimize peak demand. This problem has been previously formalized as the Peak Demand Minimization (PDM) problem and has been studied extensively [1][2][3][4][5][6][7][8]. The basic formulation of the problem is as follows: Each job j is non-preemptible, meaning once it begins execution, it must run to completion without any interruptions.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Scheduling Non-preemptible Jobs to Minimize Peak Demand

Yaw¹,

Mumey²

2017

Preprint

Self Cite

View full text Add to dashboard Cite

This paper examines an important problem in smart grid energy scheduling; peaks in power demand are proportionally more expensive to generate and provision for. The issue is exacerbated in local microgrids that do not benefit from the aggregate smoothing experienced by large grids. Demand-side scheduling can reduce these peaks by taking advantage of the fact that there is often flexibility in job start times. We focus attention on the case where the jobs are non-preemptible, meaning once started, they run to completion. The associated optimization problem is called the Peak Demand Minimization problem and has been previously shown to be NP-hard. Our results include an optimal fixed-parameter tractable algorithm, a polynomial-time approximation algorithm, as well as an effective heuristic that can also be used in an online setting of the problem. Simulation results show these methods can reduce peak demand by up to 50% versus on-demand scheduling for household power jobs.

show abstract