Games and mechanisms for networked systems: Incentives and algorithms

This chapter presents resource allocation mechanisms which take into account strategic behavior of users in wireless networks. The mechanisms include auctions and pricing schemes which differ in their allocation and communication structures, and model a large class of interference functions and interference-based coupled user utility functions. The convex optimization problems arising in interference coupled multi-carrier systems are addressed using pricing mechanisms. In addition, a regression learning scheme is studied as an alternate way to obtain the user preferences by the mechanism designer. As a main contribution, a three-step process is introduced for designing auction mechanisms by deriving the allocation and pricing functions. In the special case where the parameters of logarithmic user utilities are normalized to sum up to one, the resulting auction mechanism is shown to be budget balanced. Introduction As users play a more active role in strategic resource allocation decisions in wireless networks, the interaction between the individual users and network owners becomes more complex. The wireless users have the opportunity of manipulating the system by misrepresenting their private information for their own benefit. The network owner or designer, in turn, aims to design appropriate incentives and algorithms in order to achieve certain network level goals while eliciting true preferences from users [17].