Channelization problem in large scale data dissemination

In many large scale data dissemination systems, a large number of information flows must be delivered to a large number of information receivers. However because of differences in interests among receivers, not all receivers are interested in all of the information flows. Multicasting provides the opportunity to deliver a subset of the information flows to a subset of the receivers. With a limited number of multicast groups available, the channelization problem is to find an optimal mapping of information flows to a fixed number of multicast groups, and a subscription mapping of receivers to multicast groups so as to minimize a function of the total bandwidth consumed and the amount Of unwanted information received by receivers. In this paper we formally define two versions of the channelization problem and subscription problem (a subcomponent of the channelization problem). We analyze the complexity of each version of the channelization problem and show that they, are both NP-complete. We also find that the subscription problem is NP-complete when one flow can be assigned to multiple multicast groups. We also study and compare different approximation algorithms to solve the channelization problem, finding that one particular heuristic, flow-based-merge, finds good solutions over a range of problem configurations.