Incentive-compatible pricing strategies in noncooperative networks

The complexity of modern networks calls for decentralized control schemes where each user makes its control decisions independently based on some individual performance objectives. The operating points of such noncooperative networks are the Nash equilibria of the underlying control game. Nash equilibria are generically inefficient and lead to suboptimal network performance. Using routing as a control paradigm, a methodology is devised for overcoming this inefficiency based on pricing mechanisms. Assuming that the price for usage of each link is proportional to the congestion level at the link, it is shown that the provider can enforce any operating point it deems efficient by offering the capacity of the various links at discount prices. The incentive compatible discount vector is shown to be unique to the extent of a multiplicative constant, and its structure is specified explicitly. An adaptive algorithm for distributed computation of the incentive compatible discount vector is introduced. The applicability of the results in various practical scenaria is investigated by means of a prototype that implements the routing game as a Web-based game.