Database-driven Dynamic Spectrum Sharing (DSS) is the de-facto technical paradigm adopted by Federal Communications Commission for increasing spectrum efficiency, which allows licensed spectrum to be opportunistically used by secondary users. In database-driven DSS, a geo-location database administrator (DBA) maintains spectrum availability information over its service region in the form of a Radio Environment Map (REM), where the received signal strength from the primary user at every location is either directly measured via spectrum sensing or estimated via statistical spatial interpolation. Crowdsourcing-based spectrum sensing is a promising approach for periodically collecting spectrum measurements over a large geographic area but is unfortunately vulnerable to false spectrum measurements. Despite a large body of prior work on secure cooperative spectrum sensing, how to construct an accurate REM in the presence of false measurements remains an open challenge. In this paper, we introduce ST-REM, a novel spatiotemporal approach for securely constructing an REM in the presence of false spectrum measurements. Inspired by the self-label techniques developed for semi-supervised learning, ST-REM iteratively constructs an REM from a small number of spectrum measurements from trusted anchor sensors and many more measurements from mobile users. During each iteration, the DBA evaluates the trustworthiness of each measurement by jointly considering its spatial fitness with other trusted measurements and the mobile user's long-term behavior. By gradually incorporating the most trustworthy spectrum measurements, the DBA is able to construct a REM with high accuracy. Extensive simulation studies using a real spectrum measurement dataset confirm the efficacy and efficiency of ST-REM.

中文翻译：

---

安全众包无线电环境地图构建的时空方法


数据库驱动的动态频谱共享（DSS）是联邦通信委员会为提高频谱效率而采用的事实上的技术范例，它允许二级用户机会性地使用许可频谱。在数据库驱动的 DSS 中，地理位置数据库管理员 (DBA) 以无线电环境图 (REM) 的形式维护其服务区域内的频谱可用性信息，其中每个位置的主要用户接收到的信号强度可以直接表示为通过频谱感知测量或通过统计空间插值估计。基于众包的频谱感知是一种很有前途的方法，用于定期收集大范围地理区域的频谱测量结果，但不幸的是容易受到错误频谱测量的影响。尽管先前在安全协作频谱感知方面进行了大量工作，但如何在存在错误测量的情况下构建准确的 REM 仍然是一个悬而未决的挑战。在本文中，我们介绍了 ST-REM，这是一种新颖的时空方法，用于在存在虚假频谱测量的情况下安全地构建 REM。受到为半监督学习开发的自标记技术的启发，ST-REM 根据来自可信锚定传感器的少量频谱测量值和来自移动用户的更多测量值迭代地构建 REM。在每次迭代期间，DBA 通过联合考虑每个测量的空间适应性与其他可信测量以及移动用户的长期行为来评估每个测量的可信度。通过逐步纳入最值得信赖的频谱测量，DBA 能够构建高精度的 REM。 使用真实光谱测量数据集进行的广泛模拟研究证实了 ST-REM 的功效和效率。