In this thesis we look into the problem of grouping multicast receivers into multicast groups. More groups will give the receivers a stream better customised to their own bandwidth conditions. Less groups will better take advantage of the multicast benefits, and stress the senders and the network less.
Multicast groups also need maintaining. A big part of this is multicast congestion control. The thesis starts with a presentation of the different multicast congestion control techniques suggested by current science.
The research goal is to find the best group configuration of a given network topology. To find the best multicast configuration, genetic programming is used. An initial set of simulations is done and each simulation is evaluated. The best simulations then make the foundation for the next set of simulations. The thesis uses BRITE topology generator to generate topologies, Network Simulator (NS2) to simulate and Genetic Programming Kernel (GPC++) for the genetic evolution.