Optimizing Primary User Privacy in Spectrum Sharing Systems

Abstract

Spectrum regulators are pursuing centralized, dynamic sharing systems that will enable spectrum access for new wireless technologies. These sharing systems will leverage cognitive radio concepts to automatically identify suitable spectrum for users. Collected user information may be considered sensitive, and some incumbents are hesitant about spectrum sharing, citing privacy concerns. Privacy preserving strategies are needed to promote widespread spectrum sharing. However, privacy preserving techniques typically come at the expense of spectrum efficiency, resulting in reduced utility for the users. In this work we study this privacy-performance tradeoff. We develop a generalized spectrum sharing system architecture and formulate the multi-utility, user privacy optimization problem, where privacy is measured by exposure to potential adversary inference attacks. We derive the optimal solution for this spectrum sharing privacy problem and then formulate an efficient heuristic strategy that exploits the problem structure. Via numerical analysis, we demonstrate substantial improvement over the prevailing obfuscation strategies applied in the literature, with up to a 50% increase in privacy and negligible impact on spectrum efficiency for a real-world use case. To our knowledge, this is the first work to formally derive the optimal solution to the user privacy problem in a generalized spectrum sharing framework.