Article Photo

Fabián E. Bustamante, Tito Homem-de-Mello, and Karen Smilowitz

C3: Car-to-car cooperation

Three young faculty members take aim at traffic

Drivers in major urban areas are used to the headaches caused by traffic congestion and delays. But what if transportation is your business?

Traffic delays are more than an inconvenience to the transportation industry - they're a financial disaster. The Chicago area is the busiest transportation hub in the country, with $572 billion in freight moving through annually. Traffic delays and congestion in the area cost $4 billion each year.

A new project at McCormick joins faculty from across disciplines to try to reduce that problem. Karen Smilowitz, assistant professor of industrial engineering and management science; Tito Homem-de-Mello, associate professor of industrial engineering and management science; and Fabián E. Bustamante, assistant professor of electrical engineering and computer science, are combining their knowledge of transportation management, optimization, and distributed systems to create "car-to-car cooperation" - or C3.

"It started because the three of us are friends outside of our research," Smilowitz says. "On some level we knew what each of us was doing, but we were looking for ways to collaborate."

Each member of the team brings a unique perspective and background - Bustamante in distributed systems, Homem-de-Mello in optimization under uncertainty, and Smilowitz in routing models for freight operations - but all three say they have enjoyed learning more about the other disciplines. "When Fabián talks about networks, it's completely different from what I think about networks," Smilowitz says. "Some of the methodology we use is the same, so you can start to see where the problems are similar across disciplines."

In the beginning stages of the project, understanding each other's disciplines proved to be a challenge. Terms like drayage (the movement of containers by truck) and MANET (model ad-hoc network) required explanation. "We had to go through a whole phase of translation," Bustamante says. "We had to develop a common language, but for us it was easier because we are friends."

Simulating traffic behavior

The C3 project explores distributed systems issues in large-scale intervehicle networks and addresses a critical problem in the transportation industry: the lack of effective ways to determine routes and forecast traffic conditions based on real-time information. Traffic information is currently gathered with sensors on major thoroughfares, requiring significant investment in infrastructure equipment. Once that information is gathered, it is sent to a centralized system for analysis. While this data helps detect congestion on major routes, it isn't an effective method to predict future traffic patterns or monitor other routes such as city streets or small highways.

The C3 project is trying to leverage the computation and networking capabilities increasingly common in new vehicles to track real-time conditions, communicate with other vehicles, and then predict optimal routes. The result would be a traffic advisory system with up-to-date, no-cost information on traffic conditions, as well as better information and increased flexibility for transportation dispatchers. And because the data is not fed into one centralized source, the system could be scaled to a large number of vehicles and still be fairly reliable.

It isn't feasible to start with a large-scale implementation of a project of this scope, so Bustamante and graduate student Dave Choffnes began with the development of an integrated network and traffic simulator. They found that existing tools were good at either simulating computer networks or traffic but not both.

"I work in large-scale distributed systems," Bustamante explains, "so the idea of distributed applications built on large numbers of instrumented cars seemed to be a novel one." He adds that Smilowitz and Homem-de-Mello "know what's happening when it comes to network optimization."

Using the simulator, the group can explore traffic behavior based on many changing factors. That data is used to develop the optimization models needed to predict ideal routes. Eventually, cars with computerized instrumentation will provide information for the simulator. "The idea is that the simulator receives these measurements and calibrates itself so that it becomes more and more accurate, and we can be sure that we have good solutions," says Homem-de-Mello.

Exploring other applications

The project is more feasible given the recent surge of computerized components in cars. Navigation systems are available in a wide variety of vehicles, but the majority of these systems work using static maps, not real-time conditions. Even high-end models that alert drivers to congestion are limited to reports from major routes. Nonetheless, this trend of outfitting vehicles with computerized systems is promising for the C3 project. "We will start to see this technology trickling down to lower-end cars so that all cars will have it," Bustamante says. "We want to leverage that technology."

After the initial simulation was up and running, undergraduate students at McCormick worked to create the test nodes needed for vehicles to collect data. Moving forward, more nodes will be used to gather data. The City of Chicago has agreed to equip their fleet of vehicles for the project, providing an ideal group of vehicles to gather information from around the city.

Consumers are sure to be interested in any product that may save travel time, but the project provides many potential opportunities for transportation managers. "One of the long-term possibilities is being able to swap loads based on traffic conditions," Smilowitz says. "It would allow the possibility to swap information about loads and tasks and provide more information for the dispatchers."

For Bustamante, the benefits of a car-to-car network go far beyond transportation management. The system would create a large-scale backbone for communications unlike today's Internet or traditional wireless models. "People are increasingly concerned with the stability of the Internet," he says. "This creates a wireless environment that is completely free of infrastructure. It's more resilient, dynamic, and scalable but also more complex than current approaches."

Such a system could also provide another means for distributing information in times of crisis. "If you have some disruption, this system could be an alternative way of sending messages throughout a city," Homem-de-Mello explains.

The team is seeking funding to expand on their research, but they're not waiting on it to move ahead. "We're very excited about this project and have decided that we will work on this. People will realize how interesting the problem is, and that it makes sense to invest in the idea given its enormous potential impact," Bustamante says.