Trading Privacy for Utility in Database-Assisted Dynamic Spectrum Access
Trading Privacy for Utility in Database-Assisted Dynamic Spectrum Access
复制标题
DOI：
10.1109/tccn.2019.2919731
复制
发表时间：
2019-05
影响因子：
8.6
通讯作者：
Ahmed M. Salama;Ming Li;Loukas Lazos;Yong Xiao;M. Krunz
中科院分区：
文献类型：
--
作者：
Ahmed M. Salama;Ming Li;Loukas Lazos;Yong Xiao;M. Krunz
在动态频谱接入中，商业运营的数据库服务器经常被用来帮助机会用户(OU)查询和访问现有用户(IU)的频谱空缺。由于可能会泄露有关IU的敏感操作详细信息，特别是它们的位置，因此此查询和回答过程引入了严重的隐私问题。现有的隐私保护机制，如空间混淆、k-匿名和密码学方法，可以用于隐藏Iu位置。然而，它们不能保证由OU活动造成的干扰水平保持在所需的限制之内。此外，隐私机制在其设计中没有考虑频谱效率(即，OU传输机会的损失)。在文献中还没有系统地研究隐私、干扰和效用之间的复杂的三方权衡。在本文中，我们通过使用包含在隔离区(EZ)中的隐私区(PZ)的概念来形式化隐私和频谱效率之间的关系来解决这一挑战。在PZ内，Iu位置不可区分，而EZ保证位于PZ内的任何Iu的干扰限制。通过确定隐私或干扰要求，我们分别建立和求解了在固定和概率OU位置部署下最大化效用的相应优化问题。我们的配方可以满足多个国际单位的隐私要求。使用真实世界映射/参数的仿真结果表明，在给定对IU的固定允许干扰的情况下，Iu的私密性随着OU效用的降低而增加。此外，我们还证明了本文的性能优于巴赫拉克等人的工作。
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.