Title :
Lumpability and nearly-lumpability in hierarchical queueing networks
Author_Institution :
Informatik IV, Dortmund Univ., Germany
Abstract :
Hierarchical extended queueing networks (HQNs) have been proposed recently as a model class which allows a very efficient analysis based on an exploitation of the hierarchical model structure in the analysis of the underlying Markov chain (MC). Furthermore the hierarchical structure of the model enables the definition of symmetries due to identical subnets on model level and the automatic generation of a reduced MC resulting from exact or ordinary lumpability. In this paper the notion of symmetries is extended to nearly symmetries at model and submodel level resulting in nearly lumpability of the underlying MC. Bounds on performance quantities for nearly lumpable Markov chains are computed using bounded aggregation in the stationary case
Keywords :
Markov processes; directed graphs; performance evaluation; queueing theory; bounded aggregation; hierarchical extended queueing networks; hierarchical model structure; lumpability; model level; nearly-lumpability; performance quantities; submodel level; underlying Markov chain; Algorithm design and analysis; Computer networks; High performance computing; Intelligent networks; Performance analysis; Petri nets; Queueing analysis; State-space methods; Stochastic processes; Symmetric matrices;
Conference_Titel :
Computer Performance and Dependability Symposium, 1995. Proceedings., International
Conference_Location :
Erlangen
Print_ISBN :
0-8186-7059-2
DOI :
10.1109/IPDS.1995.395814