Security by Design via Radio Fingerprinting for Autonomous Vehicle (AV) Networks

随着新的5G服务的部署，安大略和加拿大的许多关键基础设施，如联网车辆和智能物联网结构将部署在基于RF的网络上。因此，社会将严重依赖于保护这些网络以及整个无线电频谱的能力。基于AI的发射机指纹识别等解决方案，用于识别和跟踪干扰源或恶意行为者，将成为满足下一代无线网络需求的几项关键技术之一。虽然车辆连接和自动化作为经济驱动力具有巨大潜力，但基础设施的安全性，隐私性和弹性仍然存在漏洞。该项目旨在引入新的无线电指纹识别解决方案，以提高联网和自动驾驶车辆的安全性，从而通过设计在CAV网络的物理层引入安全性，以识别和表征传输信号的独特特征。作为一个广泛研究的概念，以确保无线通信系统，信道和物理条件的影响，以及计算复杂度仍然是要解决的问题，在无线电指纹。尽管存在分析网络流量和/或应用/服务层的解决方案，但在物理层提供的安全性（和安全性）仍然是有待解决的挑战。考虑到这一点，该项目将为RF信号的特征工程和有效鉴别器的设计开发解决方案，并解决无线电指纹设计中的环境和信道条件。该项目开发的解决方案将建立在深度学习模型和基于对抗机器学习的方法之上。一旦完成，该项目将通过在CAV网络中引入无线电指纹来推进现有的CAV网络安全工作。凭借多元化和包容性的研究团队，该项目将通过物理层的身份验证和访问控制为此类解决方案引入一个非凡的机会。通过关注CAV，该项目的成果将提高安大略和加拿大私营部门的生产力和公共部门的效率，通过设计开发CAV的安全连接。

With the deployment of new 5G services, many of the Critical infrastructures in Ontario and Canada such as Connected Vehicles and Smart Infrastructures will be deployed on RF-based networks. As such, society will be heavily dependent on the ability to protect these networks as well as the Radio Spectrum as a whole. Solutions such as AI based transmitter fingerprinting to identify and track interference sources or malicious actors will be one of the several key technologies to meet the needs of next generation wireless networks. While vehicular connectivity and autonomy has significant potential as an economic driver, security, privacy and resilience of the infrastructures remain vulnerabilities. This project aims to introduce novel radio fingerprinting solutions to improve the security of connected and autonomous vehicles so that security is introduced by design at the physical layer of a CAV network to identify and characterize unique characteristics of the transmitted signals. As a widely investigated concept to secure wireless communication systems, the impact of the channel and physical conditions, as well as computational complexity remain issues to be addressed in radio fingerprinting. Although there exist solutions to analyze network traffic and/or application/service layer, security (and safety) offered at the physical layer remains challenges to be addressed. With this in mind, this project will develop solutions for feature engineering and design of effective discriminants on RF signals, and for Addressing the environmental and channel conditions in the radio fingerprinting design. Solutions developed by the project will build on deep learning models and adversarial machine learning-based methods. Once completed, the project will advance the existing efforts to secure CAV networks by introducing radio fingerprinting into CAV network. With a diverse and inclusive team of researchers, this project will introduce a remarkable opportunity for such solution through authentication and access control at the physical layer. By focusing on CAVs, the outcomes of this project will improve the productivity of the private sector and efficiency of the public sector in Ontario and in Canada by enabling the development of secure connectivity for CAVs by design.