Imperial College London

ProfessorPeterHarrison

Faculty of EngineeringDepartment of Computing

Emeritus Professor in Mathematical Modelling
 
 
 
//

Contact

 

+44 (0)20 7594 8363p.harrison Website

 
 
//

Location

 

353Huxley BuildingSouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@inproceedings{Casale:2019:10.1145/3374888.3374900,
author = {Casale, G and Harrison, P and Wai, Hong O},
doi = {10.1145/3374888.3374900},
publisher = {ACM},
title = {Novel solutions for closed queueing networks with load-dependent stations},
url = {http://dx.doi.org/10.1145/3374888.3374900},
year = {2019}
}

RIS format (EndNote, RefMan)

TY  - CPAPER
AB - Load-dependent closed queueing networks are difficult toapproximate since their analysis requires to consider state-dependent service demands. Commonly employed evaluationtechniques, such as mean-value analysis, are not equallyefficient in the load-dependent setting, where mean queue-lengths are insufficient alone to recursively determine themodel equilibrium performance.In this paper, we contribute to addressing this problem byobtaining novel solutions for the normalizing constant of stateprobabilities in the load-dependent setting. For single-classload-dependent models, we provide the first explicit exactformula for the normalizing constant that applies to modelswith arbitrary load-dependent rates, while retainingO(1)complexity with respect to the total population size. Fromthis result, we derive two novel integral forms for the normal-izing constant in multiclass load-dependent models, whichinvolve integration in the real and complex domains. Thepaper also illustrates through experiments the computationalgains and accuracy of the obtained expressions.
AU - Casale,G
AU - Harrison,P
AU - Wai,Hong O
DO - 10.1145/3374888.3374900
PB - ACM
PY - 2019///
TI - Novel solutions for closed queueing networks with load-dependent stations
UR - http://dx.doi.org/10.1145/3374888.3374900
UR - https://dl.acm.org/doi/10.1145/3374888.3374900
UR - http://hdl.handle.net/10044/1/72090
ER -