Trading Privacy for Utility in Database-Assisted Dynamic Spectrum Access
Ahmed M. Salama,Ming Li,Loukas Lazos,Yong Xiao,Marwan Krunz
DOI: https://doi.org/10.1109/tccn.2019.2919731
IF: 6.359
2019-01-01
IEEE Transactions on Cognitive Communications and Networking
Abstract:In dynamic spectrum access, commercially operated database servers are often used to assist opportunistic users (OUs) to query and access spectrum vacancies of incumbent users (IUs). This query and answer process introduces significant privacy concerns due to the potential leakage of sensitive operational details about IUs, especially their locations. Existing privacy-preserving mechanisms, such as spatial obfuscation, k-anonymity and cryptographic approaches, can be used to hide IU locations. They, however, cannot guarantee that the interference level caused by OU activity remains under required limits. Moreover, privacy mechanisms do not account for spectrum efficiency (i.e., the loss of OU transmission opportunities) in their designs. The complex three-way tradeoff between privacy, interference, and utility has not been systematically studied in the literature. In this paper, we tackle this challenge by formalizing the relationship between privacy and spectrum efficiency using the concept of a privacy zone (PZ) contained in an exclusion zone (EZ). Within the PZ, IU location is indistinguishable, whereas the EZ guarantees an interference limit for any IU located within the PZ. By fixing either the privacy or the interference requirement, we formulate and solve corresponding optimization problems that maximize utility under fixed and probabilistic OU location deployments, respectively. Our formulations can accommodate the privacy requirements of multiple IUs. Simulation results using real-world maps/parameters show that the IU’s privacy increases with decreasing OUs’ utility, given a fixed allowable interference for the IUs. Furthermore, we show that this paper outperforms the work of Bahrak et al.