Pricing Multicast Communication: A Cost-Based Approach
journal contributionposted on 23.04.2004, 00:00 by John C-I Chuang, Marvin SirbuMarvin Sirbu
Multicast and unicast traffic share and compete for network resources. To facilitate efficient and equitable resource allocation between traffic types, this paper advocates a costbased approach to multicast pricing. When prices are set to reflect actual network resource consumption, market distortion is minimized. Additionally, this paper calls for pricing multicast relative to the corresponding unicast service. This enables the end user to correctly choose multicast over unicast only when it is indeed the cheaper (and more efficient) alternative. Through the quantification of multicast link usage, this work demonstrates that the cost of a multicast tree varies at the 0.8 power of the multicast group size. This result is validated with both real and generated networks, and is robust across topological styles and network sizes. Since multicast cost can be accurately predicted given the membership size, there is strong motivation to price multicast according to membership size. Furthermore, a price ceiling should be set to account for the effect of tree saturation. This two-part tariff structure is superior to either a purely membership-based or a flat-rate pricing scheme, since it reflects the actual tree cost at all group membership levels. Explicit accounting of the control overhead allows a comparison of dense and sparse mode multicast within our cost framework. We find that sparse mode multicast maintains the 0.8 power relationship between group size and cost, while dense mode multicast is inefficient at extremely low membership levels. This suggests that, in the event when both multicast modes co-exist to serve different markets, dense mode multicast is a good candidate for flat-rate pricing and the mass-dissemination market, while sparse mode multicast is a good candidate for pricing based on membership size and the teleconferencing market.