On Probabilistic Network Calculus

Deterministic Network Calculus provides an elegant way to compute performance bounds, e.g. bounds on backlog, delay-jitter, loss ratio, and like. Such bounds are deterministic worst-case bounds. In some cases it is practically sensible that individual flows, which share a network element, are supposed statistically independent. In this case, the results of deterministic Network Calculus would yield very conservative bounds on the actual performance. In our work we show some results of Probabilistic Network Calculus that give bounds that hold in probability, but while accounting for statistical independence, yield tighter estimate of the performance.


Published in:
16th triennial conference of the Int`l Federation of Operational Research Societies; invited session on Network Calculus; Organizers: W. Seidel, B. Heidergott
Presented at:
16th triennial conference of the Int`l Federation of Operational Research Societies; invited session on Network Calculus; Organizers: W. Seidel, B. Heidergott, UK Operational Research Society, Edinburgh, UK (http://www.ifors2002.org)
Year:
2002
Laboratories:




 Record created 2004-08-31, last modified 2018-03-18


Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)