Optimizing Primary User Privacy in Spectrum Sharing Systems
Optimizing Primary User Privacy in Spectrum Sharing Systems
复制标题
DOI：
10.1109/tnet.2020.2967776
复制
发表时间：
2020-02
期刊：
影响因子：
--
通讯作者：
Matthew A. Clark;K. Psounis
中科院分区：
文献类型：
--
作者：
Matthew A. Clark;K. Psounis
频谱监管机构正在寻求集中的、动态的共享系统，这将使新的无线技术能够获得频谱。这些共享系统将利用认知无线电概念自动为用户识别合适的频谱。收集的用户信息可能被认为是敏感的，一些现任者对频谱共享犹豫不决，理由是隐私问题。需要隐私保护策略来促进广泛的频谱共享。然而，隐私保护技术通常是以牺牲频谱效率为代价的，从而导致对用户的效用降低。在这项工作中，我们研究这种隐私性能的权衡。我们开发了一个广义的频谱共享系统架构和制定的多效用，用户隐私优化问题，隐私是衡量暴露于潜在的对手推理攻击。我们推导出这个频谱共享隐私问题的最佳解决方案，然后制定一个有效的启发式策略，利用问题的结构。通过数值分析，我们证明了对文献中应用的主流混淆策略的实质性改进，隐私增加了50%，并且对真实用例的频谱效率的影响可以忽略不计。据我们所知，这是第一次正式推导出最佳解决方案的用户隐私问题的广义频谱共享框架。
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.