IEICE Information and Communication Technology Forum
Multirate Retry Loss Models Supporting Elastic Traffic of Quasi-Random Input
Ioannis Moscholios, Vassilios Vassilakis, John Vardakas, Michael Logothetis,
PDF download (255.6KB)
We propose a teletraffic loss model of a single link that accommodates multirate service-classes of elastic calls. New calls arrive in the link according to a quasi-random process (calls are generated by a finite number of traffic sources), have elastic bandwidth requirements and an exponentially distributed service time. A new call is accepted in the system with its peak-bandwidth requirement if there is available link bandwidth. If not, the call retries one or more times (single and multi-retry loss model, respectively) to be accepted in the link with reduced bandwidth. If the available link bandwidth is lower than the call’s last bandwidth requirement, the call can still try to be connected in the link by compressing its last bandwidth requirement (down to a certain bandwidth) together with the bandwidth of all in-service calls. If, after compression, the call’s bandwidth exceeds the available link bandwidth, then the call is blocked and lost. The analysis of the proposed models is based on approximate but recursive formulas, whereby we determine time and call congestion probabilities and link utilization. The accuracy of the proposed models is verified by simulation and is found to be quite satisfactory.