Spectrum broker by geo-location database

Geo-location database driven white space network is a very promising approach for improving secondary spectrum utilization. In this paper, we consider the business modeling for geo-location database driven white space network. In our proposed model, the database acts as a spectrum broker buying (reserving) bandwidth from spectrum licensees in advance, and then resells the reserved bandwidth to unlicensed white space devices (WSDs) in real-time. We study the optimal bandwidth reservation for the database with WSDs' demand uncertainty under both information symmetry and asymmetry. Under information symmetry, the database and the WSD experience the same degree of uncertainty about the market demand. We derive the optimal bandwidth reservations in a centralized/integrated manner (as a benchmark). Under information asymmetry, the WSD has more information (i.e., with less uncertainty) about demand (due to the proximity to end-users). We propose a contract-based bandwidth reservation mechanism, which ensures WSDs share their local information with the database credibly. We further characterize the optimal bandwidth reservation contract systematically. Simulations show that under information asymmetry, the optimal bandwidth reservation contract improves both the database's profit and the social welfare significantly (larger than 30% in our simulations) without sacrificing the WSDs' benefits, comparing to those mechanisms without information sharing.