Differential location privacy for crowdsourced spectrum sensing

Abstract

Dynamic spectrum access (DSA) enables secondary users (SUs) to access the underutilized licensed spectrum when the primary users (PUs) are absent and is a key solution to address the worldwide spectrum scarcity and improve the spectrum utilization. Database-driven DSA is a popular DSA paradigm and has been approved by FCC. In a database-driven DSA system, a spectrum service provider (SPP) accepts registrations from PUs and estimates the spectrum availability. To improve the accuracy of spectrum estimation in such a system, crowdsourced spectrum sensing (CSS) has been proposed, where the SPP recruits a large number of ubiquitous mobile users and outsources spectrum-sensing tasks to them. Although showing great potential, CSS requires the SPP to know the locations of mobile users for spectrum-sensing task allocation, which presents a serious privacy concern. In this paper, we propose an approach for protecting location privacy of mobile users while ensuring the sensing performance in CCS. The proposed approach is based on differential privacy and geocast, which allows the SSP to allocate tasks to mobile users without inferring their individual location information. Analytic models and task allocation strategies are developed in this paper to balance privacy, utility, and system overhead in CSS. Experimentation results based on real-world datasets show that the proposed approach can provide rigorous privacy protection to mobile users while providing effective services with low system overhead.