Performance evaluation of ``kernel based tree'' protocol

Multicasting messages to group members is now allowed thanks to many network level multicast routing protocols. Actually, current multicast routing protocols suffer from inflexibility in relation to their environment resources, and inappropriateness towards dynamic properties of group members. Thus, proposing implementations of new protocols in order to evaluate their performance is a costly issue, and it is better to simulate their behaviour before deciding whether they are efficient or not. We present a global ``adaptive multicast routing protocol{''}, based on a special multicast tree, KBT (Kernel Based Tree). This strategy may lead to a new ``QoS multicast routing protocol{''}. Our approach is based on the choice of a ``kernel{''} of ``interception points{''} in the initial multicast tree. As part of our work, we present in this paper the architecture, properties, and the main functionalities of our protocol. We finally evaluate the performance of our adaptive protocol through comparisons with classical current multicast routing protocols in the Internet, and show the enhancements provided by our solution.