Transforming Traffic Systems: The Power of Serverless Cloud Computing in Real-Time Speed Advisories

Researchers from Clemson University and the University of Alabama have developed a cutting-edge application designed to optimize traffic flow using real-time connected vehicle (CV) speed advisories. Leveraging the capabilities of Amazon Web Services (AWS), the team created a serverless cloud computing architecture to support their application, demonstrating significant improvements in traffic efficiency and safety. This innovative approach to traffic management addresses the challenges faced by traditional systems that require significant human resources and investments for installation, configuration, operation, maintenance, and upgrade.

Cloud-Based Traffic Management

The CV speed advisory application aims to minimize delays at signalized intersections by providing dynamic speed recommendations to vehicles. The system relies on AWS's serverless infrastructure, which includes services like AWS Lambda and DynamoDB, to handle the heavy computational requirements without the need for extensive on-premise infrastructure. This cloud-based approach not only ensures scalability but also reduces costs associated with traditional server maintenance and upgrades. The recent evolution of public cloud infrastructure has made it possible to support real-time CV-based Transportation Cyber-Physical Systems (TCPS) applications in the cloud, allowing public and private transportation agencies to consider replacing their on-premise roadway traffic management operations with cloud-based solutions.

Optimization Algorithms for Real-Time Advisory

The researchers developed a modular optimization algorithm for the CV speed advisory system. This algorithm divides vehicles into platoons and calculates optimal speeds to ensure smooth passage through intersections. The CV platoon assigner collects information from both traffic signals and CVs, splits the CVs into platoons based on their gap information, and computes speed advisories for the leader CVs. The CV platoon optimizer then calculates the optimal speed advisories for the follower CVs in each platoon, helping them pass the intersection while maintaining safety and operating within speed limits. The algorithm's efficiency was tested using a cloud-in-the-loop simulation setup, integrating AWS with the Simulation of Urban Mobility (SUMO) traffic simulator. This testbed was configured to simulate a 1.5-mile stretch of a four-lane highway in Clemson, South Carolina, a part of the South Carolina Connected Vehicle Testbed (SC-CVT).

Impressive Simulation Results

The results of the simulations were impressive. The serverless CV speed advisory application managed to reduce the average stopped delay at intersections by 77% and the total travel time through the corridor by 3%. Additionally, the application lowered the time-integrated time-to-collision (TIT), a measure of collision risk, by 21%. These improvements highlight the potential of using public cloud infrastructure for real-time traffic management, providing a scalable and cost-effective solution for transportation agencies. The success of this system in reducing delays and improving safety underscores the feasibility of implementing such cloud-based traffic management applications. The end-to-end delay of the application, crucial for real-time operations, averaged 452 milliseconds, well within the acceptable latency threshold of 1000 milliseconds for CV mobility applications. This low latency ensures that the speed advisories are provided to vehicles promptly, maintaining the flow of traffic and enhancing safety. The application’s ability to process data and deliver advisories within this short timeframe demonstrates its capability to operate efficiently under varying traffic conditions.

Scalable and Cost-Effective Solution

The cloud-in-the-loop simulation involved creating three different traffic density conditions: low, medium, and high. For each condition, the simulation was run multiple times with randomly generated CVs to ensure robust evaluation. The data collected from these simulations provided insights into the application's performance under different traffic scenarios. The processing time in the cloud varied within 5 milliseconds across the different traffic densities, and the average end-to-end delay varied within 20 milliseconds, indicating the application's scalability and efficiency. In addition to improving traffic flow and safety, the serverless cloud architecture offers a pay-as-you-go model, which makes it cost-effective for public transportation agencies and private companies. This model eliminates the need for significant upfront investments in infrastructure, making it an attractive option for implementing real-time CV-based roadway traffic management applications. The modular design of the optimization algorithms also ensures that the application can automatically and efficiently scale based on traffic conditions, further enhancing its practicality and usability.

Promising Future for Cloud-Based Traffic Management

The study concludes that serverless cloud architectures built upon public cloud infrastructure are capable of providing promising solutions for real-time CV-based roadway traffic management applications. Public cloud infrastructure can be considered by transportation agencies and companies to implement real-time CV-based traffic management applications, as it alleviates the need for significant investment in upgrading legacy transportation infrastructure. Future research will focus on field experiments to validate and expand the CV application developed in this study, investigating the effects of various parameters such as the number of lanes, lane widths, roadway speed limits, and CV penetration levels on the application's performance. Overall, the findings from this study suggest that cloud-based solutions can significantly improve roadway traffic conditions, reduce delays, and enhance safety, making them a viable option for modernizing traffic management systems. The successful implementation and positive results from the serverless CV speed advisory application demonstrate the potential of leveraging public cloud infrastructure to support real-time traffic management in a connected vehicle environment.