Using cloud computing to process intensive floating car data for urban traffic surveillance

Abstract

The advances in data collection techniques and geosimulation models have contributed to the abundance of geospatial data in urban systems. The surveillance of urban traffic systems relies on the effective handling of near real-time traffic observation data, which is usually data-intensive in nature. We investigated the processing of massive floating car data (FCD) for traffic surveillance in cloud-computing environments, with the goal of exploring the use of emerging cloud-computing technologies to solve data-intensive geospatial problems in urban traffic systems. The experimental results indicated that cloud-computing technologies such as Bigtable and MapReduce can provide substantial utility for data-intensive geospatial computing, as both scalability and near real-time computational performance can be adequately enhanced through our proposed data storage, management, and parallel processing models. The applicability and utility of cloud computing was evaluated for three typical geospatial computing tasks for urban traffic monitoring, namely, FCD query, FCD map matching, and speed computation for road links.