Grant: Discovery Grants Program - Individual

This Discovery project aims to develop novel solutions on the leading edge of research into the Artificial Intelligence (AI)-driven security and 5G/6G-enabled critical infrastructures from an academic perspective. Feeding into this research will be the rapid evolution of critical infrastructures such as Connected and Autonomous Vehicles (CAVs), transportation systems and utilities, and enabling technologies including AI and 5G/6G which pave the way for many new services and applications that will play a key role for securing the critical infrastructures. This project will put into practice a systems-based research agenda that is centered on integrated AI-enabled solutions for the safety and security of `5G and Beyond'-enabled critical infrastructures at all elements of the system and will form complete solutions. The research will address the following questions in the short term: 1) How can AI meet the requirements of holistic security / safety of 5G/6G-enabled critical infrastructures? 2) What communication support is needed for ultrafast exchange of data among endpoints to empower trained AI models? 3) How to decentralize and secure embedded AI models, and prevent potential bias? Successful completion of the short-term goals will lead to the long-term goal of the research, which aims to aggregate information from multiple levels and predict the sources of anomalies and/or threats in `5G and Beyond'-enabled critical infrastructures as well as the 5G/6G network. To do so, the research will develop novel explainable AI methods such as policy distillation and autonomous policy explanation.  Furthermore, autonomy that is to be enabled by extremely high data rates and sub-millisecond latencies towards 6G, introduces security vulnerabilities at multiple levels that need to be addressed by intelligent methodologies. While security and safety mechanisms can benefit from millisecond-order latency of Ultra Reliable Low Latency Communications (URLLC)-enabled 5G new radio and sub-millisecond-order latencies of the next generation URLLC (xURLLC) in 6G, it is vital to ensure that data exchanged between endpoints are not stale. With this motivation, the program will develop new reinforcement learning-based models to determine the average age of information on URLLC / xURLLC links in a critical infrastructure such as a CAV network. To ensure that the AI-driven security and safety assurance solutions are bias free, gender-based analysis+ (GBA+) analysis considerations will be done. AI models for intrusion detection or risk-analysis will be trained by providing prior knowledge as an input to ensure that decision outputs are not biased and/or security/safety is not weaponized against certain genders. Moreover, this Discovery project will expand the current diverse pool of highly qualified personnel (HQP) in Dr. Kantarci's lab to address the frontiers in these emergent fields in electrical and computer engineering, and the need for talent in Canada.