Collaborative Research: SaTC: CORE: Small: TAURUS: Towards a Unified Robust and Secure Data Driven Approach for Attack Detection in Smart Living

“智能生活”的愿景旨在提高人类的生活质量。智能家居、智能电网、智能交通或智能医疗等基石网络物理系统(CP)通过传感器驱动设备，即所谓的物联网(IoT)，生成海量的时间序列数据。此类数据可能是隐藏在良性智能生活物联网数据趋势的高随机性背后的低调隐蔽攻击的目标，从而阻碍各种应用程序依赖分析的操作的准确性。对数据分析的相互交织的依赖、错误决策的潜在民用影响以及竞争经济动机(例如，来自国家对手的动机)使得潜在的物联网和CPS领域极易受到数据完整性和可用性攻击，如创新的金牛座项目所述。这个合作项目将通过为智能生活中新兴的基于物联网的应用开发一门新的安全科学来产生巨大的影响。它解决了隐藏在由于人类行为差异导致的高度随机性背后的网络攻击和物理攻击的隐蔽攻击，并在不同攻击类型和攻击强度下制定了社区规模的统一模型。因此，金牛座项目将大幅减少同时运行的安全解决方案的数量和相应的交叉协调，以确保物联网应用的安全。此外，该项目将招募和指导本科生和研究生，包括女性和代表性不足的少数族裔学生，以及通过合作机构的各种计划培训K-12学生。金牛座项目的新颖性在于对智能生活中物联网领域的安全分析采取统一、轻量级、数据驱动的方法。基于不变量的统一异常或入侵检测理论是独一无二的，因为它捕获了来自多个物联网设备的数据中的线性和非线性关系。它还可确保在不同攻击下的大幅偏差，同时隐藏不同物联网领域在数据偏斜度、对称性、动力学和配置方面的差异。拟议的应对机制将收集有关威胁的存在、类型、严重性和战略的证据。最后，基于从证据中提取的生物信息理论概念，金牛座项目将开发一个新的统一信任框架，以在隐蔽攻击下高精度地识别受攻击的物联网设备。开发的解决方案的验证将使用来自智能电网中的智能电表和相量测量单元的真实数据集，以及智能交通中的车辆到车辆和车辆到基础设施的数据。提出的安全框架可能适用于其他智能生活物联网环境，如智能家居和供水网络。一个专门的网站将在项目期结束后的两年内维护代码、模拟和真实世界的数据集。有关数据清理、准备和转换的说明将在GitHub存储库中提供。研究成果和结果将通过金牛座网站和出版物在同行评议的高质量会议和期刊上传播。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

The “smart living” vision aims to improve human quality of life. Cornerstone cyber-physical systems (CPS) like smart homes, smart grid, smart transportation, or smart healthcare generate voluminous amount of time series data through sensor-actuator devices, the so-called Internet of Things (IoT). Such data may be a target of low-profile stealthy attacks hiding behind high randomness of benign smart living IoT data trends, thereby thwarting the accuracy of analytics dependent operations of various applications. The intertwined dependence on data analytics, potential civilian impact of wrong decisions and competitive economic motivations (e.g., by nation adversaries) make the underlying IoT and CPS domains extremely vulnerable to data integrity and availability attacks as addressed in the innovative TAURUS project. This collaborative project will create a tremendous impact by developing a new science of security for emerging IoT-based applications in smart living. It addresses stealthy attacks from both cyber and physical exploits hiding behind high randomness due to human behavioral differences and codifies a unified model at community scale under various attack types and strengths. Thus, the TAURUS project will drastically reduce the number of concurrently running security solutions and corresponding cross coordination to secure IoT applications. Additionally, the project will recruit and mentor undergraduate and graduate students, including women and underrepresented minority students, as well as train K-12 students through various schemes at partner institutions.The novelty of TAURUS project lies in the unified, lightweight, data-driven approaches towards security analytics across IoT domains in smart living. The invariant-based unified anomaly or intrusion detection theory is unique as it captures both linear and non-linear relationships in data from multiple IoT devices. It also ensures sharp deviations under various attacks yet remaining undisturbed under no attacks, while hiding differences in data skewness, symmetry, dynamics, and configuration across different IoT domains. The proposed response mechanism will gather evidence on the presence, type, severity, and strategies of threats. Finally, based on biological information theoretic concepts extracted from the evidence, the TAURUS project will develop a novel unified trust framework to identify compromised IoT devices with high accuracy under stealthy attacks. Validation of the developed solutions will use real datasets from smart meters and phasor measurement units in smart grid, and vehicle-to-vehicle and vehicle-to-infrastructure data in smart transportation. The proposed security framework is potentially applicable to other smart living IoT context such as smart homes and water distribution networks. A dedicated website will maintain codes, simulation and real-world datasets for two years beyond the project period. Instructions on data cleaning, preparation, and transformations will be available in a GitHub repository. Research findings and results will be disseminated via TAURUS website and publications in peer-reviewed high quality conferences and journals.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.