Collaborative Research: Optimal Sensor Selection and Robust Traffic Detection and Estimation in a World of Connected Vehicles

Reliable traffic management strategies require accurate knowledge of traffic levels on roads. Though the emergence of connected vehicles (CV) offers tremendous potential for sharing traffic data about vehicles' locations and speeds through wireless communications, there are both privacy concerns and bandwidth constraints - not all users want to share and not all vehicles are able to share.

This project will address both issues by designing methods to guide the selection of some road users for data sharing and analysis to provide accurate estimation of traffic levels in real time, while addressing privacy and bandwidth issues. Throughout this project, training modules in traffic and machine learning sciences will be designed at both UT San Antonio and UT Austin and students from underrepresented groups will be recruited at UTSA where 58 percent of enrolled students are minorities.

The project will: (i) consider privacy of user data hence maintaining anonymity of vehicles and users, (ii) identify sudden changes in traffic conditions due to accidents, (iii) design a time-varying selection of traffic data collected in real-time from CVs, and (iv) quantify limits on the network bandwidth and uncertainty in traffic conditions and road properties. The project's major contribution lies in advancing the use of CVs as real-time, mobile traffic sensors.

This involves the integration of concepts from multiple disciplines: traffic flow, networked systems, estimation, and machine learning theories. Specifically, the project will investigate computationally scalable methods that traffic operators can utilize to optimally sample data from CVs while satisfying privacy and bandwidth constraints, thereby monitoring traffic in real-time.

The theoretical foundations will be validated with realistic traffic setups through collaborations with the cities of Austin and San Antonio. The broader impact of the research transcends traffic networks: the computational algorithms will be applicable to related problems involving networked systems of partial differential equations and moving sensing platforms such as environmental monitoring by robot and unmanned aerial vehicles.

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.