SaTC: CORE: Medium: Physically Unclonable Wireless Systems (PUWS) for RF Fingerprinting and Physical Layer Security

这个项目的目标是研究无线通信系统利用一种新的混沌天线阵列，包括有意的，但不可克隆的随机在其几何。由于这些随机性，由混沌天线阵列传输的无线信号表现出独特且明显的指纹，可以被无线通信系统利用，以确保物联网等新兴应用的安全性。混沌天线阵列的增强指纹将用于两个目的：验证加入无线网络的无线设备，以及降低窃听者理解/解释在合法用户和接入点之间传输的无线信号的能力。本课题提出的认证和安全方法是基于硬件的，并补充了现有的基于软件的方法，如密码和加密，以在无线系统中实现更强的安全措施。该项目将开发各种配置的混沌天线阵列，研究它们在身份验证、无线通信安全和数据速率方面的性能。机器学习将用于实现准确和有弹性的身份验证。利用无掩模激光增强直接打印增材制造技术的灵活性，提出了混沌天线阵列，以引入具有空间变化的天线单元特定相位误差。所提出的阵列架构在不知道自身错误的情况下，将为基于射频指纹的身份验证和物理层安全性提供双重支持。机器学习方法，特别是深度神经网络，将用于使用增强的射频指纹进行身份验证，与使用常规天线阵列的射频指纹相比，这应该产生更高的准确性。为了从数学上分析认证能力的提高，将研究认证成功率和攻击成功率的信息理论界限。对混沌天线阵列和防御策略可能的对抗性机器学习攻击也将进行研究。最后，从无线通信的角度，考虑到无线信道分集的增强和对窃听者更具挑战性的信道估计条件，分别分析了可实现的数据速率和针对窃听者的安全性改进。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The goal of this project is to investigate wireless communication systems utilizing a new chaotic antenna array that includes intentional but unclonable randomizations in its geometry. Due to these randomizations, the wireless signals transmitted by the chaotic antenna arrays exhibit unique and strongly distinct fingerprints that can be utilized by the wireless communication systems to ensure security in emerging applications such as the Internet of Things. The enhanced fingerprints of chaotic antenna arrays will be harnessed for two purposes: authenticating wireless devices for joining a wireless network, and reducing the eavesdroppers’ capability to understand/interpret wireless signals transmitted between a legitimate user and an access point. The authentication and security methods proposed in this project are hardware-based and complement the existing software-based methods such as passwords and cryptography to achieve stronger security measures in wireless systems. The project will develop various configurations of chaotic antenna arrays, investigate their performance for authentication, wireless communication security, and data rate. Machine learning will be employed to achieve accurate and resilient authentication. Chaotic antenna arrays are proposed to introduce antenna element specific phase errors with spatial variation, leveraging the flexibilities of the mask-free laser-enhanced direct print additive manufacturing technique. The proposed array architecture, without knowledge of its own errors, will provide dual support for radio-frequency fingerprinting based authentication and physical layer security. Machine learning methods, in particular deep neural networks, will be used to perform authentication using the enhanced radio-frequency fingerprints, which should yield significantly greater accuracy compared to using radio frequency fingerprints of regular antenna arrays. To mathematically analyze the improvement in the authentication capacity, information theoretic bounds for authentication success probability and attack success probability will be studied. Possible adversarial machine learning attacks to chaotic antenna arrays and defense strategies will be also investigated. Finally, from the wireless communications perspective, the achievable data rate and the security improvements against eavesdroppers will be analyzed considering the enhanced wireless channel diversity and more challenging channel estimation conditions for eavesdroppers, respectively.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.