A good idea's journey from research paper to a successful product in the real world can be long and winding, or worse, terminate in a dead end. But PeMS, or Freeway Performance Measurement System, went from one professor's concept to a powerful computer tool in record time, and is changing transportation research and planning not only in California but other states and countries.

The story of the development of PeMS illustrates how the Institute of Transportation Studies (ITS) and its affiliated centers, California Partners for Advanced Transit and Highways (PATH) and California Center for Innovative Transportation (CCIT), take a researcher's idea, develop it, deploy and test it, and then send it on its way for use in the wider world.

PeMS is the brainchild of Electrical Engineering and Computer Science Professor Emeritus, Pravin Varaiya, who published a paper in December 1997 entitled "How to measure transportation performance." In it, he suggested that PATH and Caltrans could design and build a performance measurement system that would not only help Caltrans decide what measures to take to improve freeway travel, but more effectively incorporate new intelligent technology systems. He also urged that the information be available to the public on the web: More information about how the freeway system performed, he argued, would also allow the transportation community to make better decisions about allocating precious transportation resources.

In early 1998, Hamed Benouar, then chief traffic engineer for Caltrans and more recently, the Director of CCIT, arranged to fit Varaiya into a previously scheduled annual luncheon of Caltrans' top management.

"He showed up with only a few slides," recalled Benouar. "But he really caught their attention when he said something like, 'If you don't know what your system did yesterday, how can you know what it will do today or tomorrow.'"

As Varaiya explained recently, Caltrans at that time had an impressively large system of sensors embedded in its freeways, "but the data that they were producing was not used by anybody. If they could collect the data and use it well, or process it well, they could...learn how well the system is performing in terms of delays, speeds, and historical trends."

PATH develops and tests PeMS

Caltrans liked the idea, and six months later the first PeMS contract was awarded to PATH to develop Varaiya's idea into the first generation of PeMS and test it on one freeway in Orange County.

"There was a requirement that the data should be transmitted to a single location," recalled PATH Director Alex Skabardonis. "The only Traffic Management Center that could handle this enormous amount of information then was Caltrans' District 12 in Orange County. So that is why we began there in October 1998."

The initial foray was successful. "This exercise had given us a structure, the know-how to process this information," explained Skabardonis. It also provided new contracts from Caltrans to expand the project to the rest of California's freeways where 25,000 traffic sensors—mostly loop detectors—had been implanted in the pavement more than three decades earlier.

Boost for A New Business

For PeMs, the next step along its journey involved the private sector. While UC researchers designed the algorithms and applications, a key decision was made to turn over the software development, including the maintenance of the database system to professionals.

"In contrast to the usual scheme to have grad students or post-docs do that, we decided very early that was not the way to go in order to maintain continuity, to maintain documentation, to be certain that the system would function. In this way, the reliability of PeMS was never in question," said Varaiya.

He turned to a former grad student, Karl Petty, and his new software development company, Berkeley Transportation System. Varaiya believes that decision was key to how quickly the project moved, and Petty agrees. "It didn't really make sense to have grad students buying machines, running a data base, developing software," he explained. "They're researchers, not software developers. We were able to pay the rates to find the expertise to build something like this."

Users Beyond Caltrans

"By 2002 it became quite clear that the way we were processing the data—and more importantly, the fact that we were making it available on the web for anybody to use the data—was extremely useful," Varaiya recalled. PeMS generates 15 or 16 billion gigabytes of data to users per week.

This enormous amount of data was not only helpful to Caltrans, but to transportation students and traffic analysts who mine it freely to better understand how traffic moves—or doesn't—on congested freeways.

With each new iteration, PeMS added new features. A few years after Varaiya's paper was published, PeMS was collecting and processing information from detectors in all the state's freeways, as well as incident reports from the California Highway Patrol and FastTrak data. The complex computer tool can report causes of delay, such as too much traffic, traffic accidents, or special events. It can predict travel times based on historical trends. And although it can't solve bottlenecks, it can help traffic engineers find their causes and devise solutions. It allows them to analyze any freeway segment over any time period-an hour, day, week, month, or year. PeMS even reports when its sensors aren't working right.

Randell Iwasaki, Caltrans' chief deputy director, praised the tool. "PeMS allows both department and regional agencies the capability to see how the
system is performing (on the most highly congested urban corridors) and identify
periods of congestion."

In a recent issue of Traffic Technology International, Randell Iwasaki, Caltrans' chief deputy director, praised the tool. "PeMS allows both department and regional agencies the capability to see how the system is performing (on the most highly congested urban corridors) and identify periods of congestion. With that information, staff can then disaggregate the data further to identify causes. Detection is being expanded rapidly to support system and corridor management based on performance measurement."

CCIT's role

Before the development of PeMS, when Caltrans needed traffic data on a particular section of freeway, it took awhile. Employees were sent out to the location to gather information the old way—by counting. Now, explains Benouar, a Caltrans employee familiar with PeMS can find the same information—and it will probably be more accurate—by pushing a couple of buttons.

But CCIT needed to help Caltrans employees learn how to use the tool that it had underwritten.

"We provide a little bit of hand-holding," to help them get the hang of its new tool and understand and apply its full capabilities, explained Benouar. CCIT also identifies who in Caltrans' districts needs to know what about PeMS: Directors and managers need a good overview of how the tool works, engineers want to know how to find more specific information, such as how to look for recurring bottlenecks, and planners hope to use it to track trends over long periods of time.

"CCIT is basically the interface between the customers who want to use the tool and the researchers who continue to develop new versions of PeMS," explained Benouar.

As it has grown, it has become more user-friendly for customers outside Caltrans, such as local transportation agencies and the private sector.

Skabardonis reported that a marketing employee for the San Diego Chamber of Commerce recently called asking for information about the number of vehicles coming into her city on freeways during the summer compared to the winter months. "She's not a transportation engineer, she's a marketing person. But with a little help, in five minutes on the PeMS site she had the information," he said.

In its latest incarnation, commuters can sign on and set up MyPeMS, which allows them to customize frequent freeway routes in order to check travel times and receive other useful traffic information for a particular segment of freeway. Down the road, PATH hopes to devise a system for arterial roads. "We have a prototype, but we don't yet have a continuous flow of data," explained Skabardonis. "We hope to get data from the city of Los Angeles, which has a huge data base."

And several other states, as well as the Greek capital city, Athens, have bought the PeMS system for their own freeway systems.

"PeMS has become so useful that if PeMS goes down we get emails from all over the world," said Benouar. "In fact, one way to test who is using PeMS is to shut it down and see who calls."

In the end, Pravin Varaiya's idea was transformed into a useful tool within a few years. Benouar said that is as it should be. "We're not just about creating papers here, but really about helping society. With the structure we have here, we can do that."

PDF of story

Search ALL Issues of NewsBITS