The IPERMOB System for Effective Real-Time Road Travel Time Measurement and Prediction

摘要

Accurate, real-time measurement and estimation of road travel time is considered a central problem in the design of advanced Intelligent Transportation Systems. In particular, whether eective, real-time collection of travel time measurements in a urban area is possible is, to the best of our knowledge, still an open problem. In this paper, we introduce the IPERMOB system for efficient, real-time collection of travel time measurements in urban areas through vehicular networks. We demonstrate that travel time measurements can be accurately estimated onboard GPS-equipped vehicles, and delivered to a centralized server within a few seconds by sending a single message. Furthermore, in IPERMOB locations of travel time checkpoints can be dynamically changed through software reconfiguration, thus at a very limited cost as compared to the enormous costs of, say, installing and/or changing location of automatic vehicle identification equipment. We demonstrate the effectiveness of our approach through extensive travel time collection campaigns. In particular, our technique is shown to estimate travel time with an accuracy below 1%, with two-, three-orders of magnitude savings in both communication and storage resources with respect to existing techniques based on centralized collection of GPS traces. In the last part of the paper, we further show how real-time travel time measurements can be exploited to perform accurate, short range travel time predictions in situations where existing travel time prediction approaches are challenged (e.g., in presence of traffic congestion). The effects of vehicular network penetration rate on accuracy of travel time prediction are also discussed