Dynamic Property Rights Spectrum Access: Flexible Ownership Based Spectrum Management

In this paper we develop a framework for operator competition based on short term ownership of spectrum resources and devise a dynamic spectrum access method called "Dynamic Property Rights Spectrum Access (D-Pass)". In the D-Pass model, spectrum portions are allocated to operators on a short term basis (few sessions) by a SPS (Spectrum Policy Server) that serves as a controller/enforcer as well as a clearinghouse for spectrum allocations. Prior to each short term allocation, the SPS optimally determines a specific partition of spectrum resources among the operators to maximize a system related objective function. The operators are charged by the SPS for the amount of spectrum they are allocated. Given the spectrum allocation, the operators compete with each other for users present in the system through demand responsive pricing, in the form of an iterative bidding scheme reminiscent of simultaneous ascending auctions. At every iteration, the operators make rate and price offers for each user considering the bandwidth allocated to them and their costs associated with serving the users. The users respond by declaring the probabilities with which they will accept the service offers made. We consider two different objective functions for the SPS to maximize while determining the exact partition of the spectrum resources: (1) the total expected bandwidth utilization, (2) the minimum acceptance probability that a user accepts the offered service. We demonstrate several tradeoffs between these objectives through numerical experiments and illustrate the effect of bandwidth cost on these tradeoffs. Our results also suggest that employing short term allocation of resources could potentially lead to performance gains as opposed to static allocation of resources, especially in regimes where the bandwidth is relatively expensive.