Trust-aware Crowdsourcing for Radio Environment Mapping with Privacy Preservation

Abstract

Database-driven Dynamic Spectrum Sharing (DSS) enhances active monitoring of spectrum usage, such that secondary users are allowed to opportunistically access licenced spectrum with the no-harmful-interference-to-primary-user constraint. A geo-location database administrator (DBA) maintains spectrum availability over its service region by periodically collecting and processing spatio-temporal spectrum data to construct Radio Environment Maps (REMs). To attract mobile users in the outsourced sensing tasks of spectrum measurements, crowdsourced spectrum sensing has great potential in REM construction. However, false measurements submitted by faulty/malicious users cannot be neglected. Constructing better REMs based on crowdsourced spectrum sensing remains an open challenge, especially considering privacy protection of mobile users. In this work, we introduce TAC-REM (Trust-aware Crowdsourced spectrum sensing for the resulting REM optimization), a novel differentially-private reverse auction mechanism. TAC-REM allows the DBA to iteratively select trustworthy spectrum measurements in the presence of false measurements. During each iteration, the DBA employs a spatio-temporal scheme for trustworthiness evaluation, which acts as a guide for user recruitment in crowdsourced spectrum sensing. The proposed adaptive greedy algorithm gradually incorporates the most trustworthy spectrum measurements. The process of user recruitment can be modeled as a stochastic submodular maximization problem, while guaranteeing budget feasibility, approximate REM accuracy maximization, differential bid privacy. Extensive simulations confirm the efficacy and efficiency of the proposed mechanism.