This thesis assumed that multicast session has already been established in MPLS network. Since this MPLS multicast network supports dynamic multicast connection, we have engineered the situation where multiple new joining requests for multicast session come to the existing multicast session and ingress LER make admission decision for them from egress LER.
In our PBAC algorithm, first step is that we consider both the absolute guarantee as a guaranteed QoS policy and the probabilistic guarantee as a tolerant QoS policy. Second step is as follows. In the case of guaranteed QoS policy, we accept requests for GS multicast session and then perform GS probe process for each requests and we determine the final admission decision. In the case of tolerant QoS policy, we accept requests for GS, LSS, and BS multicast session but apply the concept of threshold between GS and LSS multicast session and different priority for them. In threshold method, we give threshold requests of LSS multicast session with low priority, while no limit for requests of GS multicast session. After determining the number of requests for GS and LSS multicast session, we perform probe process according to multicast session such GS, LSS, and BS. Next, we make final admission decision and data transmission will be done from sender to new receiver of multicast session.
The numerical analysis contains the following: queueing analysis of GS multicast session in guaranteed QoS policy in view of call and packet lavel, respectively, performance behavior for threshold, priority queueing analysis in tolerant QoS policy in view of call and packet level, respectively. In a call level, we get the blocking probability to examine the effect of probe process. In a packet level, we examine the performance evaluation after performing PBAC.
The performance results contain the following: In guaranteed QoS policy, we can get blocking probability of link and then blocking probability of request, considering the ...