Start Date: 08/16/2009
End Date: 08/15/2011

Increasing population, rising fuel costs, and stricter environmental policies have all contributed to the greater demand for improved roadway capacity and efficiency. A 2007 national study has found that inefficiencies within the roadway system commonly associated to traffic congestions have costed the U.S. $78 billion in productivity loss and 2.9 billion gallons of wasted fuel [2]. A key component in improving the efficiency of the roadway network is the availability of an advanced traffic information system to provide real-time monitoring and long-term evaluation of the roadway conditions. To that endeavor, we have developed the Highway Operation Monitoring and Evaluation System (HOMES), a traffic visualization system that allows for browsing the spatial-temporal dimension hierarchy via integrated data cube operations. The Virginia Department of Transportation (VDOT) currently utilizes HOMES to monitor traffic incidents, analyze roadway behaviors, generate operation strategies, and develop/verify highway designs. In general, HOMES can identify traffic patterns, rules, and anomalies to achieve the following benefits: efficient roadway designs; objective evaluation metrics of traffic policies; improved management of operations and emergent events; and better utilization of the roadway network.

HOMES currently monitors highways I-66 and I-395 within the Washington Metropolitan Area. It collects streaming information (arrival rate = 1 observation/sensor/min) from approximately 850 radar sensors and loop detectors. An important challenge in the development of HOMES is the provision of real-time query processing for both current and historical data under various cube operations. To address this issue, HOMES employs a spatial data warehouse approach as the underlying traffic data management strategy [3]. Fast cube-based query response times (less than 5 sec. average) are achieved by maintaining concurrent sets of aggregated and non-aggregated sensor information. Quick traffic data updates are accomplished by an incremental approach for computing the updated aggregate and non-aggregate representations of the traffic data streams.

To further enhance the query response times, we employed a caching strategy to reduce the operation costs associated with database communication and query processing. Caching allows for the reduction of resultant data in network transit and the number of direct and redundant accesses to the database. Because accessing data in main memory (i.e., data with higher locality) is more efficient (i.e., lower response times) than accessing a remotely located database, some of the traffic data will be temporarily stored in the system’s application memory pool for future recurrent accesses. A critical consideration for the caching strategy is the use of an appropriate replacement policy that adapts well to the system’s data request behavior. The Least Recently Used (LRU) has been adapted as the replacement policy in the HOMES caching scheme. As a result, HOMES allows traffic personnel to monitor and assess highway conditions in real-time while supporting efficient data operations for long-term traffic analysis.

Grant Institution: Virginia Department of Transportation

Amount: 118,070

People associated with this grant:

Chang-Tien Lu