Multicast Pricing


Since first introduced by Deering and the first audiocast experiment by IETF, multicasting is considered a very promising network service. The main reason is that it enables the efficient transmission of data addressed to a group of receivers permitting only one copy of data to traverse the common paths towards multiple destinations.

Effectively, multicasting is the service that benefits the most from network externalities. A larger number of multicast service customers implies larger economies of scale, where one customer benefits from the others since these contribute to the common cost and make access to an otherwise prohibitively expensive service possible. On the other hand, there may be some negative externality aspects.

Individual choices about the type and quality of the service may not be possible. Hence a customer may feel restrictions posed by his peers due to different service valuations or different technological capabilities. These special aspects make pricing for multicasting services an extremely interesting but complex issue. A pricing policy should reflect fairly these externality effects and provide the right incentives for new customers to join or old customers to leave the multicast session if economically justified.

The key technology element that makes possible the significant economy in network resources achieved with multicast, is the capability of the routers or of the switches of the network to duplicate at no cost incoming packet flows and direct the resulting identical flows to an arbitrary set of output links. Hence the network gains by forwarding information destined to multiple receivers over paths that have the largest possible common prefixes, since over these only a single copy of the data needs to be handled by the network. Note that an inefficient network could always use traditional unicast technology to support a multicast service. Clearly, such a technology would lead to unsustainable prices since a competitor using multicasting technology would offer lower prices due to the lower transmission cost.

However, these significant resource savings come at a certain cost: increased complexity. Scheduling of the multicast packets at the routers, routing of the packets inside the network, addressing, congestion control, quality of service issues related to multicast applications such as reliable transmission and variable transmission rates, become difficult tasks that are still under investigation. Furthermore, the decision may depend on the various assumptions made about the semantics of the multicast service supported, which in many cases are not precisely defined. The optimal solutions of some of the fundamental multicast issues such the construction of the least cost multicast tree, are related with very difficult NP-complete or even NP-hard problems, or can not at all be achieved.

As far as pricing is concerned, besides the technical difficulties and the unclear space, there are also some very important questions that still need to be answered in the modeling field. Multicast cost-based pricing is strongly associated with the game theory notion of bargaining and arbitration. The members of a multicast group receive a specific network service where cost is partly common for everyone and should be shared among them. How does one split this cost? If some customers obtain a bigger value by using the service, should they also pay a larger fraction of the common cost? This may well be the case since low value customers will leave if confronted with an equal share of the cost producing negative net benefit. Which pricing mechanisms will make such users reveal their true utility? A multicast service may be offered in an uncertain and dynamically changing environment. For instance, the number and identity of the receivers may not be known a priori, and this number may fluctuate during the multicast session. How should one price such a service, and reduce the risk of customers paying exceedingly high prices due to small participation, or the risk of the provider when prices were a priori fixed? If the service can be offered at various quality levels but only one will be actually selected reflecting, say, the worst capability of any of the receivers, how should one price such a receiver? How could one propagate the right incentives for such a receiver to leave if such an act will greatly increase the value of the service to all other receivers? These are some examples that justify the complexity of designing a sound pricing policy, and the diversity of the issues that must be addressed.

Important Publications in this area

Cost Sharing

Congestion Control



Algorithmic Aspects of Game Theory (by Christos Papadimitriou)

Topics on the border of CS, Game theory, and Economics (by Noam Nisan)