Linear and Nonlinear Pricing for Network Games With Complete and Incomplete Information

104 p.Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2007.This dissertation addresses optimal linear and nonlinear pricing policy design for a monopolistic network service provider with various types of public and private information on user types. In the communication network pricing literature, it is the linear pricing schemes that have been largely adopted, and here we investigate both linear and nonlinear pricing within the framework of a hierarchical Stackelberg (leader-follower) game, where the service provider sets prices for the resources (bandwidth) he offers as the leader, and the users respond by their choices of the amount of bandwidth (flow) they are willing to pay for. At the lower level, the presence of congestion cost (negative network effect) in the net utility functions of users leads to a noncooperative game among themselves, with Nash or Bayesian equilibrium being natural candidates for a solution. In the nonlinear pricing case, the approach is to view the problem as a reverse Stackelberg game, which is also an incentive-design problem. We also consider, for both linear and nonlinear pricing, three different scenarios based on the information sharing structure for all parties on the users' true types, namely complete information, partially incomplete information, and totally incomplete information. For each case, we obtain the optimal, or near-optimal, pricing policy that maximizes the service provider's profit given the noncooperative price-taking behavior of the users, generally for the asymptotic case regarding the number of users, since our focus is on communication networks with a large user population. Comparative studies between linear and nonlinear pricing, as well as between the three classes of informational scenarios, are carried out to evaluate profit improvement by adoption of nonlinear pricing and the service provider's game preferences.U of I OnlyRestricted to the U of I community idenfinitely during batch ingest of legacy ETD