1636 Thematic-Based Group Communication Raymond Pardede Budapest University of Technology and Economics, Hungary Budapest University of Technology and Economics, Hungary Budapest University of Technology and Economics, Hungary Copyright © 2008, IGI Global, distributing in print or electronic forms without written permission of IGI Global is prohibited. INTRODUCTION In the network level computer group-communication (one-to-many) technology, IP-multicast contributes scalability by replicating identical packets at routers. However, since its initial proposal the IP-multicast has not been widely enabled due to the slow pace of its deployment. Hence, the application-level multicast concept emerged to solve this deployment is- sue by shifting the multicast support from routers to end-systems. The article reviews the most important facts of the Application-Level Multicast and its proposed models. Furthermore, the article describes a novel concept of modeling relative density of members called bunched mode and a proposed host-end multicast transport pro- tocol called . The bunched mode is based on the , which means that it is a typical multicast scenario where there are a lot of interested hosts in certain institutes and these institutes are relatively far from each other. The developed analysis tool and the imple- mentation of the TMC called are also presented as the tools of this research. APPLICATION LEVEL MULTICAST Currently there is a fast increasing need for scalable multicast is theoretically optimal for such purposes. It can be realized in the Data-link Level, IP level and Transport/Application level (Hosszú, 2005). However, the IP Multicast has a slow deployment; it has been implemented in the most operating systems (OS) and routers, but not widely enabled. That is why the end-host based multicast is emerging, in which each member host duplicates and forwards packets. It’s simply shift- ing the multicast support from routers to end systems, see Figure 1. That solution is called application-level . As we can see from Figure 1, ALM is easy to deploy, can be understood since as a tradeoff, ALM faces the latency problems. Therefore, several ALM models are proposed. ALM model encompasses ALM routing algo- rithm, ALM protocol, ALM topology, and so forth. For designing an ALM model, the metric or good- ness of the generic ALM model must be recognized. The goodness of the ALM system can be measured by some parameters, such as control overhead, robustness of the overlay, stress, and stretch. Control overhead means the ratio of the necessary control messages sent by the clients to each other and other word, the control overhead is a metric to exam- ine the scalability of the overlay to large groups. Each member on the overlay exchanges refresh messages with all its peers on the overlay. Those messages build the control overheads at different routers, different links and different members of the multicast group. For ef- should be low (Banerjee et al., 2002). Robustness of the overlay of the ALM protocols is measured by quantifying the extent of the disruption in data delivery when various members fail, and the time it takes for the protocol to restore delivery to the other members. Since hosts are potentially less stable than routers, it is important for ALM protocols to decrease the effect of host failures.