Preserving Geo-Indistinguishability of the Primary User in Dynamic Spectrum Sharing

Dynamic spectrum sharing (DSS) has been taken as an encouraging method to address the increasing demand for wireless spectrum resources. Without privacy protection in location, however, the primary user will hesitate to share spectrum with secondary users. In this paper, we present LpriDSS, the first scheme for a spectrum administrator to select spectrum-sharing secondary users while implementing the geo-indistinguishability of primary user. First, we formulate the secondary users selection without privacy in DSS system as an auction problem and demonstrate the location inference attacks based on the winner sets in detail. Then, a ranking metric to characterize the administrators' preference for secondary users is defined. Moreover, a probability of each secondary user being selected as a winner is calculated and used during the process of secondary user selection. Finally, a truthful payment computation is designed according to that probability. Thorough theoretical analysis and simulation studies show that LpriDSS can simultaneously achieve geo-indistinguishability, approximate social welfare maximization, and truthfulness.