Skip to main content
Open Access Publications from the University of California

Freeway Performance Measurement System (PeMS)


The freeway Performance Measurement System (PeMS) collects real time traffic data from sensors and generates performance measures of vehicle miles traveled, hours traveled, and travel time. This project is sponsored by the California Department of Transportation (Caltrans). PeMS provides tools and reports for traffic planners, operators, and engineers. It has a Web interface. Growing traffic demand in metropolitan areas has far outpaced increases in freeway lane-miles in the United States. The solution to congestion lies in increasing the efficiency of existing infrastructure. Performance measurement is the first step in effective management and operation of any system. Currently, the freeway system is not managed scientifically. Planning and operating decisions are made without accurate knowledge of the performance of each part of the system. PeMS collects data from automatic sensors that are already installed on most of California freeways. Its large database of real time and historical data 2 allows us to accurately measure the performance of freeways and its trends. Traffic planners need this information to allocate the available resources to improve mobility. PeMS computes performance measures and other traffic quantities from sensor data. Among them are speed, vehicle-hours of delay, vehicle-miles traveled, and travel time statistics. These values can be visualized in plots and summarized in reports, and they are available online through a Web interface. Policy makers can use PeMS to evaluate the effect of their decisions and set performance targets, planners monitor trends in congestion and respond with congestion-reduction measures, engineers view detailed data to improve conditions at specific locations, and travelers use the information to make more informed decisions. Researchers use PeMS's database to analyze traffic behavior on a large scale. We present some results from studies on freeway capacity, travel time variability, and the impact of incident on overall delay. In these cases, using observations from a large number of locations and times allows us to characterize traffic flow statistically. PeMS processes raw data into useful forms. It computes speed from single loop detectors, predict travel time from real time and historical data, and detect and fix data errors. We describe these data processing algorithms, which are based on empirical models and fitted to historical data.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View