A novel analytical framework compounding statistical traffic modeling and aggregate-level service curve disciplines: network performance and efficiency implications

This paper demonstrates that higher network resource efficiency can be achieved by using resource management protocols which consider service disciplines based on service curves together with statistical traffic modeling. To this end, an appropriate analytical framework is introduced which allows calculation of the performance statistically guaranteed to any flow out of an aggregate. This feature enables the analytical framework to be applied to the elements of the core network where aggregates of traffic are considered instead of single flows in order to avoid scalability problems. Given that flows are modeled in the analytical framework through switched batch Bernoulli processes (SBBPs), the whole queueing system is denoted as SBBP/Sc/1/K. The performance is calculated in terms of loss probability and delay distribution. The proposed framework is applied in a significant multinode case study.