Priority-based Network Services

Although today’s Internet predominantly uses the flat-fee pricing scheme, the need for more sophisticated pricing has grown, especially for domains where bandwidth resources are scarce and expensive. This paper focuses on how to determine the price vector that maximizes the system revenue in priority-based networks. Similar problems have been examined using equilibrium analysis, but the assumptions needed to produce a static equilibrium may not hold in some realistic network environments. We propose a new method that does not require such assumptions. First, we have developed a method for computing the expected revenue for a given price vector. Then, instead of exhaustively trying out all the possible price vectors, we reduce the search space based on a set of pruning rules and apply a heuristic search to the remaining search space to find the near-optimal price vector. We demonstrate that the proposed scheme finds a near-optimal price vector efficiently, and compare our method with the equilibrium analysis. Acknowledgements: The first author is partially supported by Rutgers Research Resource Council Grant.