A Cross-Layer Approach to Taming Cyber-Physical Uncertainties in Vehicular Wireless Networking and Platoon Control

Transforming the traditional, single-vehicle-based safety and efficiency control, next-generation vehicles are expected to form platoons for optimizing roadway usage and fuel efficiency while ensuring transportation safety. Two basic enablers of vehicle platooning are vehicular wireless networking and platoon control. Tightly coupled with the physical process of vehicle movement and wireless signal propagation, however, vehicular wireless networking and platoon control are subject to and challenged by complex dynamics and uncertainties in both the physical domain and the cyber domain. To address the challenges of cyberphysical uncertainties, we propose a cross-layer framework where wireless networking and platoon control interact with each other to tame cyber-physical uncertainties:

• Based on the real-time capacity region of wireless networking and the physical process of vehicle movement, platoon control selects its control strategy and the corresponding requirements on the timeliness and throughput of wireless data delivery to optimize control performance (e.g., maximizing roadway utilization while ensuring safety);
• Based on the requirements from platoon control, wireless networking controls co-channel interference and adapts to cyberphysical uncertainties to ensure the timeliness and throughput of single-hop as well as multi-hop broadcast; for proactively addressing the impact of vehicle mobility on wireless broadcast, wireless networking also leverages input from platoon control on vehicle movement prediction. With access to the USDOT/SAIC Connected Vehicle Technology (CVT) testbed and with extensive expertise in wireless networking, control engineering, and mathematics, the PIs will investigate networking and control mechanisms to realize the above cross-layer framework, thus enabling predictable, provable qualities in vehicular wireless networking and platoon control. The proposed research makes novel, significant contributions to the model building, algorithm design, system analysis, and architecture for taming cyber-physical uncertainties in cooperative vehicle operation:
• The physical-ratio-K (PRK) wireless interference model integrates protocol model’s locality with physical model’s high- fidelity, thus enabling agile, predictable interference control in the presence of cyber-physical uncertainties; Together with our control-theoretic approach to parameter identification, the PRK model also addresses the challenge of identifying the set of links that interfere with each link, thus filling the gap in the theory of pairwise-interference-model-based scheduling.
• Signal-map-based protocol signaling and design addresses the physical challenges of large interference range, anisotropic, asymmetric wireless communication, and the collision of broadcast-receiver-feedback;
• Virtual broadcast backbone addresses the challenges that vehicle mobility poses to real-time multi-hop broadcast, and the PIs also leverage the physical model of vehicle mobility in protocol design;
• Networked control with random topology switching and time delay serves as a new framework, not only for the specific application of platoon control, but also for the control theory as a discipline and for stochastic differential equations as a mathematical subject;
• The study on the interplay between the frequency of topology switching and the speed of networked control will introduce new methods for treating multi-timescale structures and will open up new avenues for studying multi-timescale stochastic systems;
• The cross-layer framework enables the joint optimization of wireless networking and platoon control in taming cyberphysical uncertainties, and the mathematical tools (e.g., stochastic approximation, switching ordinary differential equations, and switching diffusions) serve as a foundation for reasoning about the jointly-optimized wireless networking and platoon control.

By effectively addressing cyber-physical uncertainties in cooperative vehicle operation, these contributions are significant because they are expected to enable predictable, provable qualities in vehicular wireless networking and platoon control (e.g., the timeliness of sensing and the optimality of control), thus laying a solid foundation for safe, efficient transportation with networked vehicles. The models and algorithms generated by the proposed research will also have a positive impact on other domains of wireless networked sensing and control such as unmanned aerial vehicles and smart power grids.